{"paper_id":"2503.13950","title":"A Note on the Asymptotic Properties of the GLS Estimator in Multivariate Regression with Heteroskedastic and Autocorrelated Errors","authors":["Koichiro Moriya","Akihiko Noda"],"paper_date":"2025-03-18","paper_type":"theoretical","paper_url":"https://arxiv.org/pdf/2503.13950v1","page_url":"https://thequant.space/flowcharts/a-note-on-the-asymptotic-properties-of-the-gls-estimator-in/","scores":{"empirical_rigor":3,"math_complexity":8.5,"quadrant":"Lab Rats"},"proposes":{"status":"available","statement":"The paper proposes Wald test statistics based on feasible GLS estimators (Prais‑Winston and Cochrane‑Orcutt types) that explicitly accommodate heteroskedasticity and autocorrelation in multivariate regression errors.","quote":"In this paper, we study the asymptotic properties of the GLS estimator for the multivarite parametric regression models and derive Wald statistics based on the Prais-Winston (PW) and Cochrane-Orcutt (CO) type feasible GLS estimators. These Wald statistics are explicitly de-veloped to accommodate heteroskedasticity and autocorrelation in error terms.","summary":"We study the asymptotic properties of the GLS estimator in multivariate regression with heteroskedastic and autocorrelated errors. We derive Wald statistics for linear restrictions and assess their performance. The statistics remains robust to heteroskedasticity and autocorrelation.","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Error terms are heteroskedastic and autocorrelated.","type":"statistical","stated":"explicit","quote":"These Wald statistics are explicitly de-veloped to accommodate heteroskedasticity and autocorrelation in error terms.","support":"full"},{"assumption":"The GLS estimator is analyzed under the asymptotic framework with T → ∞.","type":"statistical","stated":"explicit","quote":"2 Model and Test In this section, we analyze the asymptotic properties of the GLS estimator in multiple regres-sion with heteroskedastic and autocorrelated errors.","support":"full"},{"assumption":"Assumptions 1, 2 and 3 (as defined in the paper) hold for the asymptotic results.","type":"model","stated":"explicit","quote":"Suppose that Assumptions 1, 2 and 3 are satisﬁed.","support":"full"},{"assumption":"In the simulation, error terms follow either a heteroskedastic process or a heteroskedastic VAR(1) autocorrelated process.","type":"statistical","stated":"explicit","quote":"We examine two cases for the error terms: (i) heteroskedastic error terms and (ii) heteroskedastic and autocorrelated error terms. In the latter case, we assume that the autocorrelated errors follow a VAR(1) process.","support":"full"}]},"datasets":{"status":"available","items":[{"name":"Synthetic simulation data","provider":null,"kind":"synthetic","access":"unknown","period":null,"universe":null,"quote":"In our simulation, σ2 i is drawn from the uniform distribution U(0.5, 1), and ρ is set to 0.3. The intercept α0 = [α1, ..., αN]′ is set to α0 = 0N,1 under the null hypothesis, while under the alternative hypothesis, it is speciﬁed as α1 = 0.1 and αj = 0 for j = 2, . . . , N. We consider the following values for T, N, and k: T ∈{200, 400, 800, 1600, 3200}; k ∈{3, 5}; and N ∈ 6 {6, 25}.","support":"full"}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"synthetic"},"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":"none","findings":[{"finding":"The proposed Wald tests maintain accurate size (empirical rejection rates close to nominal significance levels) under both heteroskedastic and autocorrelated error settings, and performance improves with larger sample sizes.","kind":"statistical","quote":"In simulation experiments, we conﬁrm that our proposed Wald tests perform well regardless of the presence of heteroskedasticity and autocorrelation. The results show that our proposed Wald tests exhibit accurate size properties, with empirical sizes converging to the nominal levels as the sample size increases, regardless of the number of portfolios.","support":"full"},{"finding":"Both the original and modified GRS tests exhibit severe size distortions (over‑rejection) when errors are autocorrelated, regardless of the number of cross‑sectional units.","kind":"negative","quote":"In contrast, the original and modiﬁed GRS tests exhibit severe Type I errors when error terms are autocorrelated, and the HAR Wald test tends to over-reject the null hypothesis as the number of cross-sectional units increases. Both the original and modiﬁed GRS tests exhibit substantial size distortions relative to the normal error case.","support":"full"},{"finding":"The HAR Wald test over‑rejects the null hypothesis for larger cross‑sectional dimensions (N = 25), indicating reduced reliability as the number of portfolios increases.","kind":"negative","quote":"Furthermore, while the HAR Wald test performs well for N = 6, it severely over-rejects the null hypothesis for N = 25, indicating reduced reliability as the number of portfolios increases.","support":"full"},{"finding":"Overall, the proposed Wald statistics are robust to relaxed error‑term assumptions, outperforming existing tests in finite‑sample simulations.","kind":"theoretical","quote":"Therefore, we conclude that the proposed Wald statistics have desirable properties even when the assumptions about the error terms are relaxed. We also constuct Wald statistics for linear restrictions on the estimator and assess their ﬁnite-sample performance.","support":"full"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":21,"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":2,"extracted_at":"2026-11-14T19:51:30Z"},"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/"}