{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,17]],"date-time":"2026-10-17T02:40:24Z","timestamp":1792204824280,"version":"4.4.0"},"reference-count":33,"publisher":"Oxford University Press (OUP)","issue":"4","license":[{"start":{"date-parts":[[2026,9,17]],"date-time":"2026-09-17T00:00:00Z","timestamp":1789603200000},"content-version":"vor","delay-in-days":259,"URL":"https:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,10,7]]},"abstract":"<jats:title>Summary<\/jats:title>\n                  <jats:p>Recent advances in deep learning have highlighted the phenomenon of benign overfitting in overparameterized statistical models, sparking significant interest in understanding its foundations. Owing to its simplicity and practical relevance, the ordinary least-squares interpolator has become a key object of study for gaining theoretical insight into this phenomenon. While the properties of ordinary least squares are well understood in classical underparameterized settings, its behaviour in the overparameterized regime, unlike that of ridge regression or the lasso, remains comparatively less explored. We contribute to this growing literature by deriving new algebraic and statistical results for the minimum $ \\ell_{2} $-norm ordinary least-squares interpolator. In contrast to much of the existing work, which focuses on prediction risk, we centre our analysis on parameter estimation and inference, which are fundamental to many statistics and causal inference applications. Specifically, we establish overparameterized analogues of (i) the leave-$ k $-out formulas, (ii) the omitted-variable bias formula and (iii) the Frisch\u2013Waugh\u2013Lovell theorem. Under the Gauss\u2013Markov model, we further extend the Gauss\u2013Markov theorem and analyse variance estimation under homoscedasticity in the overparameterized setting. Collectively,these results provide a systematic framework for studying parameter estimation and inference in overparameterized linear models, offering a novel perspective on benign overfitting beyond its implications for prediction.<\/jats:p>","DOI":"10.1093\/biomet\/asag039","type":"journal-article","created":{"date-parts":[[2026,6,19]],"date-time":"2026-06-19T11:53:50Z","timestamp":1781870030000},"source":"Crossref","is-referenced-by-count":0,"title":["Benign overfitting beyond prediction: the ordinary least-squares interpolator"],"prefix":"10.1093","volume":"113","author":[{"suffix":"By","given":"Dennis","family":"Shen","sequence":"first","affiliation":[{"name":"Department of Data Sciences & Operations, University of Southern California , 3670 Trousdale Parkway , Los Angeles, California 90089 . dennis.shen@marshall.usc.edu","place":["U.S.A"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Dogyoon","family":"Song","sequence":"additional","affiliation":[{"name":"Department of Statistics, University of California, Davis , One Shields Avenue , Davis, California 95616 . dgsong@ucdavis.edu","place":["U.S.A"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2704-2353","authenticated-orcid":false,"given":"Peng","family":"Ding","sequence":"additional","affiliation":[{"name":"Department of Statistics, University of California, Berkeley , 425 Evans Hall , Berkeley, California 94720","place":["U.S.A"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jasjeet","family":"Sekhon","sequence":"additional","affiliation":[{"name":"Google DeepMind , 1600 Amphitheatre Parkway, Mountain View , California 94043, . jas@jsekhon.com","place":["U.S.A"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"286","published-online":{"date-parts":[[2026,9,17]]},"reference":[{"key":"2026101621420722400_asag039-B1","doi-asserted-by":"crossref","first-page":"42","DOI":"10.1017\/S0370164600014346","article-title":"IV.\u2014On least squares and linear combination of observations","volume":"55","author":"Aitken","year":"1936","journal-title":"Proc. R. Soc. Edin"},{"key":"2026101621420722400_asag039-B2","doi-asserted-by":"crossref","DOI":"10.2307\/j.ctvcm4j72","volume-title":"Mostly Harmless Econometrics: An Empiricist\u2019s Companion","author":"Angrist","year":"2008"},{"key":"2026101621420722400_asag039-B3","doi-asserted-by":"crossref","first-page":"486","DOI":"10.1214\/20-AOS1965","article-title":"Predictive inference with the Jackknife+","volume":"49","author":"Barber","year":"2021","journal-title":"Ann. Statist."},{"key":"2026101621420722400_asag039-B4","doi-asserted-by":"crossref","first-page":"30063","DOI":"10.1073\/pnas.1907378117","article-title":"Benign overfitting in linear regression","volume":"117","author":"Bartlett","year":"2020","journal-title":"Proc. Nat. Acad. Sci"},{"key":"2026101621420722400_asag039-B5","doi-asserted-by":"crossref","first-page":"203","DOI":"10.1017\/S0962492921000039","article-title":"Fit without fear: remarkable mathematical phenomena of deep learning through the prism of interpolation","volume":"30","author":"Belkin","year":"2021","journal-title":"Acta Numerica"},{"key":"2026101621420722400_asag039-B6","doi-asserted-by":"crossref","first-page":"1167","DOI":"10.1137\/20M1336072","article-title":"Two models of double descent for weak features","volume":"2","author":"Belkin","year":"2020","journal-title":"SIAM J. Math. Data Sci"},{"key":"2026101621420722400_asag039-B7","doi-asserted-by":"crossref","first-page":"1503","DOI":"10.3982\/ECTA20289","article-title":"Exact bias correction for linear adjustment of randomized controlled trials","volume":"92","author":"Chang","year":"2024","journal-title":"Econometrica"},{"key":"2026101621420722400_asag039-B8","author":"Cheng","year":"2026"},{"key":"2026101621420722400_asag039-B9","doi-asserted-by":"crossref","first-page":"755","DOI":"10.1093\/biomet\/asm046","article-title":"On a generalization of a result of W. G. Cochran","volume":"94","author":"Cox","year":"2007","journal-title":"Biometrika"},{"key":"2026101621420722400_asag039-B10","doi-asserted-by":"crossref","DOI":"10.1201\/9781003484080","volume-title":"A First Course in Causal Inference","author":"Ding","year":"2024","edition":"1st edn"},{"key":"2026101621420722400_asag039-B9997508","author":"Ding","journal-title":"Linear Model and Extensions"},{"key":"2026101621420722400_asag039-B11","volume-title":"Statistical Methods for Research Workers","author":"Fisher","year":"1925"},{"key":"2026101621420722400_asag039-B12","doi-asserted-by":"crossref","first-page":"180","DOI":"10.1016\/j.aam.2006.12.003","article-title":"On regression adjustments to experimental data","volume":"40","author":"Freedman","year":"2008","journal-title":"Adv. Appl. Math"},{"key":"2026101621420722400_asag039-B13","doi-asserted-by":"crossref","first-page":"387","DOI":"10.2307\/1907330","article-title":"Partial time regressions as compared with individual trends","volume":"1","author":"Frisch","year":"1933","journal-title":"Econometrica"},{"key":"2026101621420722400_asag039-B15","doi-asserted-by":"crossref","first-page":"949","DOI":"10.1214\/21-AOS2133","article-title":"Surprises in high-dimensional ridgeless least squares interpolation","volume":"50","author":"Hastie","year":"2022","journal-title":"Ann. Statist"},{"key":"2026101621420722400_asag039-B14","doi-asserted-by":"crossref","DOI":"10.1007\/978-0-387-21606-5","volume-title":"The Elements of Statistical Learning","author":"Hastie","year":"2001"},{"key":"2026101621420722400_asag039-B16","doi-asserted-by":"crossref","first-page":"295","DOI":"10.1214\/aos\/1009210544","article-title":"On the distribution of the largest eigenvalue in principal components analysis","volume":"29","author":"Johnstone","year":"2001","journal-title":"Ann. Statist."},{"key":"2026101621420722400_asag039-B17","doi-asserted-by":"crossref","first-page":"95","DOI":"10.1007\/s00440-016-0754-9","article-title":"On the impact of predictor geometry on the performance on high-dimensional ridge-regularized generalized robust regression estimators","volume":"170","author":"Karoui","year":"2018","journal-title":"Prob. Theory Rel. Fields"},{"key":"2026101621420722400_asag039-B18","doi-asserted-by":"crossref","first-page":"14557","DOI":"10.1073\/pnas.1307842110","article-title":"On robust regression with high-dimensional predictors","volume":"110","author":"Karoui","year":"2013","journal-title":"Proc. Nat. Acad. Sci"},{"key":"2026101621420722400_asag039-B19","doi-asserted-by":"crossref","first-page":"815","DOI":"10.1093\/biomet\/asaa103","article-title":"Regression adjustment in completely randomized experiments with a diverging number of covariates","volume":"108","author":"Lei","year":"2021","journal-title":"Biometrika"},{"key":"2026101621420722400_asag039-B20","doi-asserted-by":"crossref","first-page":"295","DOI":"10.1214\/12-AOAS583","article-title":"Agnostic notes on regression adjustments to experimental data: reexamining Freedman\u2019s critique","volume":"7","author":"Lin","year":"2013","journal-title":"Ann. Appl. Statist"},{"key":"2026101621420722400_asag039-B21","doi-asserted-by":"crossref","first-page":"993","DOI":"10.1080\/01621459.1963.10480682","article-title":"Seasonal adjustment of economic time series and multiple regression analysis","volume":"58","author":"Lovell","year":"1963","journal-title":"J. Am. Statist. Assoc"},{"key":"2026101621420722400_asag039-B22","doi-asserted-by":"crossref","first-page":"822","DOI":"10.1214\/25-AOS2489","article-title":"The generalization error of max-margin linear classifiers: benign overfitting and high dimensional asymptotics in the overparametrized regime","volume":"53","author":"Montanari","year":"2025","journal-title":"Ann. Statist."},{"key":"2026101621420722400_asag039-B23","doi-asserted-by":"crossref","first-page":"965","DOI":"10.1111\/rssb.12374","article-title":"A scalable estimate of the out-of-sample prediction error via approximate leave-one-out cross-validation","volume":"82","author":"Rad","year":"2020","journal-title":"J. R. Statist. Soc. B"},{"key":"2026101621420722400_asag039-B24","doi-asserted-by":"crossref","first-page":"812","DOI":"10.1214\/12-AOS982","article-title":"Estimation in high-dimensional linear models with deterministic design matrices","volume":"40","author":"Shao","year":"2012","journal-title":"Ann. Statist."},{"key":"2026101621420722400_asag039-B25","doi-asserted-by":"crossref","first-page":"2125","DOI":"10.3982\/ECTA21248","article-title":"Same root different leaves: time series and cross-sectional methods in panel data","volume":"91","author":"Shen","year":"2023","journal-title":"Econometrica"},{"key":"2026101621420722400_asag039-B26","first-page":"63642","volume-title":"Proc. 37th Int. Conf. Neural Info. Proces. Syst","author":"Spiess","year":"2023"},{"key":"2026101621420722400_asag039-B27","doi-asserted-by":"crossref","first-page":"14516","DOI":"10.1073\/pnas.1810420116","article-title":"A modern maximum-likelihood theory for high-dimensional logistic regression","volume":"116","author":"Sur","year":"2019","journal-title":"Proc. Nat. Acad. Sci"},{"key":"2026101621420722400_asag039-B28","first-page":"1","article-title":"Benign overfitting in ridge regression","volume":"24","author":"Tsigler","year":"2020","journal-title":"J. Mach. Learn. Res"},{"key":"2026101621420722400_asag039-B30","first-page":"5228","volume-title":"Proc. 35th Int. Conf. Mach. Learn (Proc. Mach. Learn. Res. 80)","author":"Wang","year":"2018"},{"key":"2026101621420722400_asag039-B31","author":"Zhang","year":"2017"},{"key":"2026101621420722400_asag039-B32","doi-asserted-by":"crossref","first-page":"1401","DOI":"10.1214\/21-AOS2156","article-title":"Ridge regression revisited: Debiasing, thresholding and bootstrap","volume":"50","author":"Zhang","year":"2022","journal-title":"Ann. Statist."},{"key":"2026101621420722400_asag039-B33","doi-asserted-by":"crossref","first-page":"327","DOI":"10.1093\/jrsssb\/qkad006","article-title":"Debiased and thresholded ridge regression for linear models with heteroskedastic and correlated errors","volume":"85","author":"Zhang","year":"2023","journal-title":"J. R. Statist. Soc. B"}],"container-title":["Biometrika"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/biomet\/advance-article-pdf\/doi\/10.1093\/biomet\/asag039\/71188196\/asag039.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/biomet\/article-pdf\/113\/4\/asag039\/71188196\/asag039.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/biomet\/article-pdf\/113\/4\/asag039\/71188196\/asag039.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,17]],"date-time":"2026-10-17T01:42:20Z","timestamp":1792201340000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/biomet\/article\/doi\/10.1093\/biomet\/asag039\/8812157"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":33,"journal-issue":{"issue":"4","published-print":{"date-parts":[[2026,10,7]]}},"URL":"https:\/\/doi.org\/10.1093\/biomet\/asag039","relation":{},"ISSN":["0006-3444","1464-3510"],"issn-type":[{"value":"0006-3444","type":"print"},{"value":"1464-3510","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2026]]},"published":{"date-parts":[[2026]]},"article-number":"asag039"}}