{"id":"29ab75b4-8062-4d41-9e07-3ab1f1a00cea","arxiv_id":"2603.11497","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A variance estimator that adds an extra square term to cluster- and time-robust estimators is proposed to prevent over-rejection when means are heterogeneous and clusters are serially correlated.","lead":"This paper proposes a conservative variance estimator for panel data where cluster-level means differ and observations are correlated across clusters and time. The body's central result is a usable estimator with asymptotic size control, but the abstract promises impossibility and optimality results that the manuscript does not contain.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 2 hinges on Assumption 4(b), which fails for balanced panels with fixed G; the proof of conservativeness does not cover that regime.","rationale":"The reader's verdict is CONDITIONAL, and the weakest assumption identified is Assumption 4(b). My independent reading reaches the same point: the proof's bridge from Vcon >= Vadj to Vcon >= Vtrue(1+o(1)) is Proposition 2, and the only assumption enforcing negligibility of the within-cluster serial-covariance terms is Assumption 4(b). This is a real substantive restriction, not a technical artifact: in a balanced panel with a bounded number of clusters, the omitted term is of the same order as the true variance for any fixed AR(1) coefficient, so Proposition 2 fails exactly where the paper's own running example would suggest applicability. The body's central estimator may still be conservative in that regime because Vcon contains additional positive terms, but the manuscript does not prove this. The abstract's unproved impossibility and optimality claims are a separate, already-flagged issue; they do not change the conditional status of the body's result. Therefore the reader's CONDITIONAL verdict should stand without modification.","tokens_in":17557,"tokens_out":25887,"duration_ms":226251,"concrete_test":"Analytically compute W/lambda_n for the balanced panel with G fixed, one observation per cell, and Y_{g,t} = rho Y_{g,t-1} + epsilon_{g,t} with epsilon iid, |rho|<1, T -> infinity, G fixed. Here W = 2G sum_{m=1}^{T-1} (T-m) rho^m and lambda_n = G Var(sum_{t=1}^T Y_t), so W/lambda_n -> 2 rho/(1+rho) != 0, showing Assumption 4(b) fails. If the authors believe conservativeness still holds in this regime, this should be accompanied by a direct proof that Vcon - Vtrue is PSD up to o(lambda_n), without invoking Proposition 2.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The asymptotic-conservativeness claim is established by the chain Vcon >= Vadj (Prop. 1) and Vadj = Vtrue(1+o(1)) (Prop. 2). Proposition 2 depends critically on Assumption 4(b): the within-cluster-across-time covariance terms W in the last line of Eq. (14) must be o(lambda_n). This is not a consequence of the CLT conditions (Assumptions 1-3); it is a separate restriction on the growth of lambda_n relative to the sum of |N^{T cap G}_{t,g}| |N^{T cap G}_{t+m,g}| theta_{n,m}^{1-2/p}. The paper verifies it only in the balanced panel with G asyomptotically T and one observation per cell. In the same AR(1) balanced-panel model but with G fixed and T -> infinity, W/lambda_n -> 2 rho/(1+rho) > 0, so Assumption 4(b) fails even though Assumptions 1-3 can hold. In that regime Proposition 2 does not follow, and the paper supplies no alternative direct proof that Vcon - Vtrue is PSD up to o(lambda_n). Thus the central claim is established only under a substantive, unverified domain restriction.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a conservative variance estimator for sums from a triangular array with heterogeneous means and two-way cluster dependence, where one dimension has weak (ψ-) dependence. It argues that standard CHS-type plug-in estimators can be anticonservative under heterogeneous means (Example 3), and proposes adding a positive term 2Σ_t y_t y_t' to the serial-correlation component. The main results are: Theorem 1 (a CLT via KMS), Theorem 2 (consistency of V̂_con under a high-level eigenvalue condition), Proposition 1 (V_con − V_adj is PSD), and Proposition 2 (V_adj = V_true(1+o(1)) under Assumption 4). The abstract additionally advertises impossibility results, dual-cone necessary-and-sufficient conditions, and eigenvalue-truncation optimality under three criteria, none of which appear in the body. The paper includes simulations and an empirical application showing that the proposed HM standard errors control size better than existing methods.","tokens_in":17859,"tokens_out":7173,"duration_ms":63614,"significance":"If the central claims hold, the paper makes a useful contribution: it extends variance estimation to settings with both heterogeneous means and cross-cluster serial dependence, and it provides a simple modification of the CHS estimator. The use of KMS limit theory rather than exchangeability representations is a strength and permits more general DGPs. The paper also delivers concrete, falsifiable predictions in the form of Assumptions 1–4, and its numerical results support the qualitative point. However, the advertised scope in the abstract is far broader than the actual content, and the main asymptotic-conservativeness chain has a load-bearing gap in a natural fixed-G, T→∞ regime. The core idea is plausible, but the manuscript in its present form needs substantial revision.","major_comments":[{"comment":"The abstract claims (i) impossibility of consistent estimation in general, (ii) necessary and sufficient conditions for conservative estimation via dual cones, and (iii) an eigenvalue-truncation estimator that is optimal under minimal-correction, pointwise-level, and pointwise-MSE criteria. None of these results appear in Sections 1–4 or the Appendix; the paper only presents a specific consistent conservative estimator. This is not a mere framing issue: the advertised contributions are missing. Either the missing theory must be added, or the abstract must be rewritten to match the actual content.","section":"Abstract and Sections 1–4"},{"comment":"Proposition 2 relies on Assumption 4(b) to ensure that the within-cluster-over-time covariance terms in the last line of Eq. (14) are asymptotically negligible. The paper verifies Assumption 4(b) only for the running example with G≍T (Section 3, paragraph after Assumption 4). For a balanced panel with G fixed and T→∞, one observation per (g,t) cell, and AR(1) within clusters with coefficient ρ, Assumptions 1–3 can hold with λ_n≍T, but λ_n^{-1} Σ_{m≥1} Σ_{t=1}^{T-m} Σ_g |N^{T∩G}_{t,g}| |N^{T∩G}_{t+m,g}| θ^{1-2/p}_{n,m} → 2ρ/(1+ρ) > 0. Hence Assumption 4(b) fails, Lemma 2's proof (Eq. 48) does not apply, and the conclusion V_adj = V_true(1+o(1)) is not established in this regime. Consequently, the chain V_con ≥ V_adj ≈ V_true does not prove conservativeness, and the central validity claim is established only under a substantive domain restriction that is not stated in Theorem 2 or Proposit","section":"Assumption 4(b) and Proposition 2"},{"comment":"Theorem 2 states consistency of V̂_con under Assumptions 1 and 3 plus the high-level condition λ_min(V_con)/λ_n ≥ 1+o(1). This condition is not derived from the primitive assumptions; it is essentially part of the desired conclusion. The subsequent remark that the condition is automatically satisfied under Proposition 2 depends entirely on Proposition 2, which is subject to the Assumption 4(b) limitation above. As stated, Theorem 2 does not establish that the proposed estimator is consistent for a variance target that is itself conservative for the true variance in the general setting claimed. The theorem should be restated to make the required eigenvalue condition a primitive or to prove it under the stated assumptions.","section":"Theorem 2"},{"comment":"The proof of Proposition 1 (Eq. 36) uses the nonnegativity of the kernel weights ω(m,M) to conclude that the serial-correlation part of V_con − V_adj is PSD. The paper only assumes |ω(.)|≤1 (Section 3, before Theorem 2). Standard HAR kernels are typically nonnegative, but the proposition as stated applies to any kernel satisfying |ω|≤1, which is false for sign-changing kernels. Either add an explicit assumption ω(m,M)≥0 in Proposition 1 and Assumption 4, or adapt the proof to handle kernels with negative weights.","section":"Proposition 1"}],"minor_comments":[{"comment":"Assumption 1(c) is written as 'sup_n max_i ∥Y_i∥_p < ∞ a.s.' The 'a.s.' appears misplaced: the norm is a moment, not a random quantity. Probably the intended condition is a uniform p-th moment bound. Please clarify.","section":"Section 2, Assumption 1(c)"},{"comment":"The sentence 'Assumption 4(b) is not required, because the last line of Equation (14) no longer features in the variance expression as an adjustment term' is specific to the pure time-series case. In the panel setting with clusters, Assumption 4(b) is essential. The text should make this distinction explicit to avoid confusion.","section":"Section 3, after Theorem 2"},{"comment":"The simulation table reports rejection rates but no Monte Carlo standard errors or confidence intervals. Given that the HM method is a new proposal, reporting simulation uncertainty would help assess whether the differences are meaningful. Also, the heterogeneity term β^h_gt is set to an alternating ±0.1; a brief explanation of why this is a representative design would improve the exposition.","section":"Section 4.1, Table 2"},{"comment":"In the first part of the proof, the sentence 'The derivation for the time dimension is identical as N^T_{t(i)} ⊂ N^∂_n(i;0)' is terse. Since N^T_t is a cluster on the time dimension, observations in the same time period are at distance 0 by definition; spelling this out would improve readability.","section":"Proof of Lemma 1"},{"comment":"The reference to Xu and Yap (2024) is an arXiv preprint; the paper might be updated with a published version if available. Also, the reference to KMS (Kojevnikov et al., 2021) is used heavily and correctly, but the precise theorem numbers (Theorem 3.2, Theorem A1) should be cross-checked against the published version.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The abstract advertises impossibility results and optimality theory that do not appear anywhere in the manuscript. This is a serious mismatch and may indicate an incomplete submission or an overclaim; the editor should verify whether the missing sections exist. The fixed-G failure of Assumption 4(b) is a substantive gap that affects the central claim; however, it may be fixable by adding a direct proof of conservativeness or by clearly restricting the scope. The paper's core idea and numerical evidence are promising, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague, here's my read.\n\nThe paper's actual contribution is solid: a conservative variance estimator (Vcon) for two-way cluster dependence plus weak cross-cluster serial correlation, robust to arbitrary mean heterogeneity. The proof strategy is to verify conditions of the KMS network-dependence CLT and then do PSD algebra. Proposition 1's decomposition is clean, and Example 2 makes a real point: the CHS exchangeability-plus-stationary-time-effects representation rules out natural DGPs like independent AR(1) firms. Extending KMS to heterogeneous means is a legitimate contribution. The simulation, while quick, makes the point: CHS and CGM over-reject with heterogeneous means, and HM gets close to nominal when rho is high.\n\nNow the problems. The abstract advertises three big results: impossibility of consistent variance estimation, dual-cone necessary and sufficient conditions for conservativeness, and an eigenvalue-truncation estimator optimal under three criteria. None of that appears anywhere in Sections 1-4. This isn't a small mismatch; a reader cannot evaluate the advertised theorems at all. The body's theorems are limited to consistency of Vcon and PSD comparisons. Either those abstract promises need proofs, or they need to go.\n\nThe second soft spot is Assumption 4(b). The stress-test is right: in a balanced panel with fixed G and T growing, the within-cluster-across-time covariance term W in Eq. (14) is not o(lambda_n). For an AR(1), W/lambda_n -> 2rho/(1+rho). So Proposition 2's claim that Vadj = Vtrue(1+o(1)) fails in that regime. The paper only verifies Assumption 4(b) for the G asyomptotically T running example. The conclusion could still hold—if Vadj overestimates Vtrue, it only helps conservativeness—but the paper doesn't provide that direct argument. So the central proof has a real gap for fixed-G asymptotics. That's worth fixing or explicitly flagging as a limitation.\n\nMinor: no code or replication package, and the Monte Carlo has no precision info. For a theory paper that's less important, but it would help.\n\nWho is this for? Empirical microeconomists and time series people using cluster-robust inference with serially correlated time effects. The proposed estimator is easy to implement and plausibly useful. The paper deserves a serious referee, but the revision should be major: align the abstract with the content, supply or remove the promised theory, and address the Assumption 4(b) gap.\n\nRecommendation: send to peer review, not desk reject, but with clear expectations for revision.","headline":"The body gives a genuinely useful conservative variance estimator for psi-dependent panels with heterogeneous means; the abstract promises much more than the text delivers.","tokens_in":18307,"tokens_out":5184,"would_cite":true,"duration_ms":42322,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper shows that when group means are heterogeneous, standard cluster-robust and serial-correlation-robust variance estimators can understate the true variance, and proposes a simple conservative estimator that restores asymptotic size","keywords":["variance estimation","heterogeneous means","cluster dependence","serial correlation","conservative inference","two-way clustering","panel data","asymptotic size"],"falsifier":"Simulate a balanced panel with G=T=200, within-cluster AR(1) shocks with ρ=0.95, mean shifts alternating in sign across clusters and periods, and a known null; compute rejection rates of a 5% test using Vcon. If rejection rates remain clearly above 5% (say, above 10%) as G and T grow at the paper's rates, the claimed asymptotic conservativeness is violated.","tokens_in":17434,"feed_emoji":"📊","tokens_out":7596,"duration_ms":64067,"temperature":0.7,"pith_summary":"This paper addresses a blind spot in standard variance estimation: when observations have heterogeneous means—unit-specific nonzero means that cancel only in aggregate—plug-in variance estimators that replace means with sample averages target an uncentered second-moment object. Under independence this overstates the variance, but under dependence the cross terms of the means can be negative, making the estimand understate the true variance and tests over-reject. The paper proposes a simple fix: add a lag-zero sum-of-squares term to the serial-correlation-robust panel variance estimator. It proves that the resulting estimand is consistent for its own target, that this target is a positive-semidefinite expansion of the kernel-adjusted covariance, and that the kernel-adjusted covariance converges to the true variance, so tests based on the new estimator control size under weak cross-cluster dependence. A sympathetic reader would care because over-rejection from such anticonservative standard errors invalidates inference in design-based and nonstationary settings.","feed_headline":"Variance estimator stays conservative when group means differ","feed_subtitle":"It adds a simple lag-zero term to panel standard errors, restoring asymptotic size control.","key_machinery":"The estimator Vcon is formed from the usual cluster-and-time double sums of raw Y_i Y_j' terms plus kernel-weighted lagged period sums, with an added 2 sum_t y_t y_t' term at lag zero. The proof works by decomposing Vcon - Vadj into positive semidefinite pieces: outer products of within-cluster sums of means, within-time sums of means, and sums of (mean_t + mean_{t+m}) outer products. Because each piece is a sum of outer products, the difference is psd. The asymptotics are carried by a ψ-dependence framework with dependence coefficients θ_{n,s} and neighborhood-growth counts c_n(s,m;k), which control covariances of products of observations and make the variance of the estimator's error vanis","core_discovery":"The central formal claim is that the proposed estimator Vcon is consistent for its own target Vcon (Theorem 2), that Vcon exceeds the kernel-adjusted variance Vadj by a positive semidefinite matrix (Proposition 1), and that Vadj converges to the true variance Vtrue (Proposition 2). Together these imply that Vcon is asymptotically conservative for Vtrue: its smallest eigenvalue is no smaller than the true variance's smallest eigenvalue in the limit, so a normal test based on Vcon will not exceed its nominal size under the maintained assumptions. The paper also shows by example that the standard plug-in estimator for this setting can be anticonservative, with the gap driven by products of hete","pith_inferences":["The abstract advertises impossibility, dual-cone characterization, and optimality criteria for choosing among valid estimators, but the main text's theorems establish consistency and conservativeness, not a formal optimality guarantee; a reader should treat the optimality claims as programmatic rather than proven in this version.","Because conservativeness is engineered by adding a lag-zero sum-of-squares term, the estimator sacrifices power; a natural extension is a data-driven shrinkage factor that shrinks Vcon toward Vadj while preserving the positive semidefinite gap.","The guarantee depends on within-cluster serial correlation decaying fast enough relative to λ_n; in designs with near-unit-root within-cluster dynamics, the conservative correction may be too slow to appear at realistic sample sizes.","A direct empirical follow-up would compare Vcon's confidence intervals with bias-corrected or bootstrap-calibrated alternatives in the industry-portfolio setting, where the reported standard errors are substantially larger than existing methods."],"forward_implications":["In a balanced panel with weak cross-cluster serial correlation, standard errors computed from Vcon will be asymptotically at least as large as those from the conventional serial-correlation-robust panel estimator, so nominal 5% tests will not over-reject under H0.","The overestimation is bounded in simple AR(1) settings: the estimator inflates the long-run variance by at most a factor of two when serial correlation is weak, and the correction becomes negligible for local-to-unity processes.","The estimator extends to regression coefficients: when the score and squared-score arrays satisfy the assumptions, OLS coefficients remain asymptotically normal and Vcon constructed from residuals is conservative for the coefficient variance.","The method does not require estimating, differencing, or smoothing the heterogeneous mean sequence, so it applies to arbitrary patterns of mean heterogeneity.","In the independent-observation limit the dependence assumptions are automatically satisfied, so consistency is retained with no added convergence-rate penalty."],"fun_headline_variants":["Fix for panel variance when means differ and errors correlate","New variance estimator robust to heterogeneous means and serial dependence","Simple fix makes panel variance estimates conservative","Existing variance estimators can be anticonservative; new one fixes that","Adding a lag-zero term restores conservative variance estimates"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the extra correlation between observations in the same cluster at different times must shrink fast enough relative to the overall variance; if within-cluster serial correlation decays too slowly, the conservative guarantee fails.","fun_headline_variants_meta":{"raw":{"variants":["Fix for panel variance when means differ and errors correlate","New variance estimator robust to heterogeneous means and serial dependence","Simple fix makes panel variance estimates conservative","Existing variance estimators can be anticonservative; new one fixes that","Adding a lag-zero term restores conservative variance estimates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000304,"raw_usage":{"total_tokens":1520,"prompt_tokens":617,"completion_tokens":903,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":361,"completion_tokens_details":{"reasoning_tokens":827}},"tokens_in":361,"tokens_out":903,"duration_ms":8257,"temperature":1.0,"reasoning_tokens":827,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T18:20:59.664016+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate a balanced panel with G=T=200, within-cluster AR(1) shocks with ρ=0.95, mean shifts alternating in sign across clusters and periods, and a known null; compute rejection rates of a 5% test using Vcon. If rejection rates remain clearly above 5% (say, above 10%) as G and T grow at the paper's rates, the claimed asymptotic conservativeness is violated.","supporting_citations":[],"review_version":1}