{"id":"a7c8862b-060e-4b81-8e2a-1fe8742ff04c","arxiv_id":"2607.16535","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"For the local-to-zero cross-validated log-likelihood criterion CVLL_c, the expectation of the quadratic term in its Taylor expansion converges to the asymptotic mean squared error of the zero-frequency kernel spectral estimator when 4/5 < c < 1.","lead":"This paper derives the asymptotic behavior of one key term in a local cross-validation criterion for choosing the bandwidth of a kernel spectral estimator at zero frequency. It is a step toward justifying data-driven bandwidth choice for HAC standard errors used in econometrics, but it leaves other terms in the expansion unproved.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 0.1 is sound for the quadratic term, but the CVLL_c justification depends on unproved negligibility of the (0.16)/(0.17) expectations outside Gaussian white noise.","rationale":"The paper proves a precise theorem about the expectation of the quadratic term (0.18), and the proof of that theorem is largely self-contained: Lemma 0.2 handles the bias component and Lemma 0.5 handles the variance component, with the Dirichlet-kernel calculation explaining the c > 4/5 condition. I checked the main algebra of Lemma 0.5 and did not find an internal inconsistency that would invalidate Theorem 0.1 as stated. The load-bearing gap is instead the step from Theorem 0.1 to the abstract's claim of 'some justification' for CVLL_c: the expectation of the full Taylor expansion is E[CVLL_c(τ) − L~ ] = n^{4/5−c}E(0.16) − n^{4/5−c}E(0.17) + c0(τ) + o(1), where the first two terms are known to vanish only for Gaussian white noise. The authors explicitly state that they believe these terms are negligible but do not prove it. This is not an internal contradiction; the theorem is stated as a building block. But it is the single weakest link in the central claim, and the reader's CONDITIONAL verdict correctly reflects that. I would not move the verdict: the concern is real, already identified by the reader, and does not require a change in the verdict. The proposed Monte Carlo check would settle whether the unproved conjecture is true empirically for standard non-Gaussian linear processes.","tokens_in":13461,"tokens_out":34569,"duration_ms":319524,"concrete_test":"Monte Carlo simulation for a non-Gaussian linear process, e.g., AR(1) with φ=0.5 and iid standardized t_5 or centered exponential innovations, using the Parzen window. For n in {2^10, 2^12, 2^14, 2^16}, c=0.9, and τ in {0.5, 1, 2}, estimate with at least 10^4 replications the scaled expectations E[n^{4/5−c} Σ_{j=1}^{⌊n^c⌋} (fhat_j^j/f_j −1)(I_j/f_j −1)] and E[n^{4/5−c} Σ_{j=1}^{⌊n^c⌋} (I_j/f_j −1)(fhat_j^j/f_j −1)^2], using the known true f_j. If these do not converge to 0 as n grows, or if the normalized E[CVLL_c(τ) − L~ ] does not approach c0(τ), the conjecture fails outside Gaussian white noise; if they do converge, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that E[CVLL_c(τ) − L~ ] tracks the asymptotic mean squared percentage error c0(τ) requires that, after the n^{4/5−c} scaling, the expectations of terms (0.16) and (0.17) in the Taylor expansion are o(1). The paper proves this only in the Gaussian-white-noise case, where independence of {I_j} from the leave-one-out estimate makes the expectations exactly zero. For a general linear process with iid non-Gaussian innovations, periodogram ordinates are not independent, so Cov(fhat_j^j, I_j) and mixed third-order moments such as E[(I_j/f_j −1)(fhat_j^j/f_j −1)^2] are not identically zero. The text acknowledges this gap immediately after Eq. (0.19): 'We believe that it could be proved... but we do not pursue this here.' If those two scaled expectations do not vanish, then E[CVLL_c(τ) − L~ ] equals c0(τ) plus additional terms that could be of the same order, so Theorem 0.1 alone does not justify CVLL_c-based HAC bandwidth selection. This is precisely the weakest assumption identified by the reader, and it is load-bearing because the abstract's 'some justification' rests on it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the local-to-zero cross-validated log-likelihood criterion CVLL_c(τ) for bandwidth selection in kernel spectral estimation at frequency zero. The authors define CVLL_c(τ) as a leave-one-out average over Fourier frequencies j=1,...,n^c with 0<c<1, and consider a Taylor expansion of the criterion difference L~(τ)-L~. Their main theorem (Theorem 0.1) states that, for m=τ n^{1/5}, ñ=n^c, and 4/5<c<1, the scaled expectation of the quadratic term in the expansion, (1/2)n^{4/5-c} ∑ E( fhat_j^j/f_j - 1)^2, converges to the asymptotic mean squared percentage error c0(τ) of the zero-frequency spectral estimator, where c0(τ)=1/2[τκ+τ^{-4}(h^{(2)}f^{(2)}(0)/f(0))^2]. The proof decomposes the leave-one-out error and proves two lemmas: one on the squared bias of the lag-window estimate (Lemma 0.2) and one on the variance contribution of the innovation periodogram (Lemma 0.5). The paper explicitly acknowledges that the expectations of two other Taylor terms, (0.16) and (0.17), are only shown to vanish in the Gaussian white-noise case, and conjectures but does not prove their negligibility for general linear processes.","tokens_in":13797,"tokens_out":4457,"duration_ms":42785,"significance":"If the central claim were fully established, the paper would provide an important bridge between the local CVLL criterion and the classical asymptotic MSE of the Parzen spectral window estimator, thereby giving a formal justification for HAC bandwidth selection via CVLL_c. The paper correctly identifies that existing global CVLL results cannot be transplanted to the local-to-zero case because Parseval's formula fails for local sums. The authors are also commendably transparent about the unproved negligibility of the (0.16) and (0.17) terms. However, as presented, the result justifies only the quadratic term in the Taylor expansion, not the expectation of the full CVLL_c difference; the missing control of (0.16) and (0.17) is load-bearing for the advertised conclusion. The proof of Lemma 0.5 also contains an incomplete step. Thus the paper is of moderate significance pending a fix of these gaps.","major_comments":[{"comment":"The abstract claims that the results provide 'some justification' for CVLL_c-based HAC bandwidth selection. But the paper explicitly states after Eq. (0.19) that the expectations of (0.16) and (0.17) are only proved negligible in the extremely special Gaussian white-noise case, and that 'we believe that it could be proved... but we do not pursue this here.' For general non-Gaussian linear processes, leave-one-out estimates are not independent of I_j, so E[(fhat_j^j/f_j-1)(I_j/f_j-1)] and E[(I_j/f_j-1)(fhat_j^j/f_j-1)^2] need not vanish. If these terms have the same order as the quadratic term after n^{4/5-c} scaling, then E[CVLL_c(τ)-L~] would be c0(τ) plus additional unknown terms, so Theorem 0.1 alone does not justify CVLL_c-based bandwidth selection. This is the central load-bearing gap and should be addressed, either by proving the negligibility under explicit regularity conditions o","section":"After Eq. (0.19)"},{"comment":"The proof of Lemma 0.5 contains a missing line: 'and . Thus' at Eq. (0.75) obscures the simplification of E[γ^ϵ_r γ^ϵ_s]. In particular, the derivation leading to Eq. (0.77) requires a careful accounting of the nonzero fourth-moment terms when |r|=|s|≠0 and when r or s equals zero. The sentence is broken and the step is not verifiable as written. Since Lemma 0.5 is essential for the variance contribution in Theorem 0.1, this missing line and the underlying algebra should be completed and checked.","section":"Lemma 0.5, around Eq. (0.75)"},{"comment":"The expression in Eq. (0.36) is not written consistently: it should be τ^{-4}(h^{(2)} f^{(2)}(ω_j)/f(ω_j))^2, comparing with the definition of b0(τ) in Eq. (0.29). Also, Theorem 0.1 uses c0(τ) while Eq. (0.11) writes 'co(τ)'. These are presentation errors, but they make it difficult to verify the proof's algebra and should be fixed.","section":"Eq. (0.36) and surrounding notation"}],"minor_comments":[{"comment":"Reference [1] has typos ('Bandwith', 'Specturm'), and Lemma 0.3 cites 'Chen and Hurvich (1998)' while the reference list gives 'Chen, W. & Hurvich, C. (2000)'. Please standardize.","section":"References"},{"comment":"The abstract appears twice in the manuscript text (once at the top and again after the arXiv line). This duplication should be removed.","section":"Abstract"},{"comment":"The Op term in Eq. (0.15) is written as Op( n^~ (fhat_j^j/f_j-1)^3 ) with the tilde over n in a nonstandard way; please clarify the notation and the order of the remainder term.","section":"Eq. (0.15)"}],"recommendation":"major_revision","confidential_remarks":"The paper is honest about its main gap, but that gap is too central for acceptance as is. If the authors can prove or convincingly bound the omitted expectations (0.16) and (0.17) under explicit conditions, the paper could become acceptable. The proof of Lemma 0.5 also needs a small but essential repair."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Careful, honest paper, but the scope is narrower than the abstract lets on. Theorem 0.1 is a genuine local-to-zero result for the expectation of the quadratic term (0.18), and I don't see a flaw in the proof. The catch is that the paper does not actually deliver E[CVLL_c(τ)−L~] ≈ c0(τ), because the cross terms (0.16) and (0.17) are only shown to be negligible for Gaussian white noise. The authors admit this immediately after (0.19), so it's not hidden—but it's load-bearing for the claim that CVLL_c justifies HAC bandwidth selection.\n\nWhat's new: Robinson's global CVLL proof uses Parseval, which fails in the local-to-zero regime; the authors identify that and build a different argument. Lemma 0.2's bias approximation for the Parzen window is carefully done, and Lemma 0.5's variance calculation plus the Dirichlet-kernel bound is real work. The reduction of the leave-one-out estimator to the lag-window estimator via Robinson's (C.7)/(C.9) is sensible. Citations are appropriate—Robinson and Parzen are the right baselines, and the Chen–Hurvich Dirichlet bound is used transparently.\n\nThe soft spots are proportional to the claim. For a general linear process with non-Gaussian innovations, periodogram ordinates aren't independent, so expectations of (0.16) and (0.17) are not identically zero; no proof shows they are o(1) after the n^{4/5−c} scaling. If they are not, E[CVLL_c(τ)−L~] has extra same-order terms and the bandwidth-selection story collapses. The proof of Lemma 0.5 also has a missing line around (0.75), plus assorted typos that need a cleanup pass.\n\nWho gets value: econometricians and time-series statisticians working on HAC standard errors and frequency-domain cross-validation. This is a stepping-stone, not a completed theory. It deserves a serious referee—the theorem is non-trivial and I believe correct, and the open negligibility question is stated plainly. The right outcome is conditional acceptance: either prove the missing negligibility or reframe the contribution as the first-term theorem, which is still publishable. I'd cite it for the local-to-zero quadratic-form result, not for a full CVLL_c justification.","headline":"A correct-looking local-to-zero theorem for the CVLL_c quadratic term, but the bandwidth-selection justification depends on an unproved negligibility conjecture that the authors themselves flag.","tokens_in":14215,"tokens_out":5650,"would_cite":true,"duration_ms":58461,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["62M15","62M10","62G05"],"pacs":[],"model":"deepseek-v4-flash","headline":"A local cross-validated likelihood term is shown to converge to the asymptotic mean squared error of the zero-frequency spectral estimator, supporting automatic HAC bandwidth selection.","keywords":["kernel spectral estimation","bandwidth selection","cross-validated log-likelihood","HAC standard errors","long-run variance","zero frequency","asymptotic mean squared error","local-to-zero"],"falsifier":"A simulation or exact asymptotic computation for a non-Gaussian linear process (for example, an ARMA model with Student-t innovations) that evaluates E[CVLL_c(τ) − L~] and compares it with c0(τ) over τ would decide the matter. If the difference between the two does not vanish relative to n^{4/5−c} for c in (4/5, 1), then the unproved terms are not negligible and the bandwidth selector would be biased. One could also directly compute the expectations of (0.16) and (0.17) at low frequencies; if they are not o(n^{c−4/5}), the theorem's premise fails.","tokens_in":13370,"feed_emoji":"📈","tokens_out":12601,"duration_ms":115913,"temperature":0.7,"pith_summary":"This paper establishes a precise asymptotic connection between a 'local-to-zero' version of the cross-validated log-likelihood (CVLL_c) and the mean squared error (MSE) of a kernel spectral estimator at frequency zero. Specifically, it proves that for a shrinking set of Fourier frequencies (from 1 to n^c with 4/5 < c < 1), the scaled expectation of the quadratic term in the Taylor expansion of CVLL_c tends to the classical variance-plus-squared-bias formula c0(τ) = 1/2[τκ + τ^{-4}(h^{(2)} f^{(2)}(0)/f(0))^2]. Since HAC standard errors depend on an accurate estimate of f(0), this gives theoretical support for using CVLL_c to choose bandwidth automatically, though the full justification rests on an unproved conjecture that other terms in the expansion are negligible. The proof overcomes a technical obstacle: the usual Parseval identity that works for global CVLL fails for local sums, so the paper develops a Dirichlet-kernel bound that controls the local frequency sum.","feed_headline":"Local cross-validation term converges to zero-frequency MSE","feed_subtitle":"It ties a data-driven bandwidth rule to the accuracy of long-run variance estimation","key_machinery":"The central object is the Taylor expansion of the local cross-validated log-likelihood difference L~(τ) − L~, whose quadratic term (0.18) is a sum of squared relative errors of the leave-one-out estimates. The proof of Theorem 0.1 decomposes each relative error fhat_j^{(j)}/f_j − 1 into a negligible high-order term, an innovation-noise term g*_j (a centered periodogram-based spectral estimate of the innovations), and a bias term E[f~_j]/f_j − 1. Lemma 0.5 shows that the variance part converges to τκ by bounding the sum of Dirichlet kernels that appear because the frequency sum is local; this bound substitutes for the Parseval identity used in global CVLL proofs. Lemma 0.2 shows that the bias","core_discovery":"At the heart of the paper is Theorem 0.1: when the bandwidth is m = τ n^{1/5} and the number of local frequencies is n^c with 4/5 < c < 1, the scaled expectation (1/2)n^{4/5-c} Σ_{j=1}^{n^c} E[(fhat_j^{(j)}/f_j - 1)^2] converges to c0(τ). Here fhat_j^{(j)} is the leave-one-out kernel spectral estimate at Fourier frequency ω_j, f_j is the true spectrum, and c0(τ) is exactly the asymptotic MSE of the discrete periodogram average estimate of f(0): a term τκ for the variance and a term τ^{-4}(h^{(2)} f^{(2)}(0)/f(0))^2 for the squared bias. The theorem focuses on the quadratic term of the Taylor expansion of CVLL_c(τ) around the true spectrum; the paper shows that this term carries the τ-depende","pith_inferences":["A concrete next step would be to prove the negligibility of the linear and cross terms using mixing or cumulant conditions; if those terms have non-negligible expected limits, c0(τ) would gain an additive shift that changes the optimal bandwidth constant τ*, a possibility the paper does not quantify.","The Dirichlet-kernel technique suggests that other global criteria, such as AIC or FPE, might also have local-to-zero versions whose expectations match the MSE of the spectral density at a single frequency; one could test this by deriving analogous expansions for those criteria.","The paper's assumption of a finite fourth innovation moment and the appearance of the kurtosis term E[ε^4]−1 in the variance lemma (though it vanishes asymptotically) hints that finite-sample CVLL_c could be sensitive to heavy tails; a simulation study of an AR(1) with t-distributed innovations would reveal whether the bandwidth selection deteriorates in moderate samples.","Since the theorem requires only that f be twice continuously differentiable near zero and that the underlying process be short-memory with finite fourth moment, the result may carry over to fractionally integrated processes at frequency zero, though the rates would need re-derivation because the spectral density may be unbounded."],"forward_implications":["If the authors' conjecture about the negligible linear and cross terms holds, then E[CVLL_c(τ) − L~] equals c0(τ) up to a constant, so the minimizer of CVLL_c(τ) asymptotically matches the minimizer of the zero-frequency ASE, giving a fully data-driven bandwidth rule for HAC standard errors.","The theorem pins down the convergence rate: the scaling n^{4/5−c} is exactly what is needed for the quadratic sum to have a finite limit, and the condition c > 4/5 is necessary for the bias term to survive in the limit; c < 1 keeps the frequency window local to zero.","Because the result holds for each fixed τ, it can be applied to compare different bandwidth constants τ in the same local frequency window, turning CVLL_c into a practical selector that does not require knowledge of f, its derivatives, or the noise variance.","The proof's technical core—a Dirichlet-kernel bound for sums over low frequencies—is a reusable tool for other local-to-zero model selection problems where global Parseval arguments are unavailable."],"fun_headline_variants":["Local CVLL bandwidth selector matches zero-frequency MSE","Cross-validation term hits spectral estimator MSE at zero","Bandwidth rule tied to long-run variance accuracy","Local-to-zero CVLL converges to spectral MSE","Proof links local CVLL to zero-frequency error"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the expectations of the linear term (0.16) and the cross term (0.17) in the Taylor expansion of CVLL_c are negligible for general stationary processes; this is proven only for Gaussian white noise, and the authors state they believe it holds more generally but do not provide a proof.","fun_headline_variants_meta":{"raw":{"variants":["Local CVLL bandwidth selector matches zero-frequency MSE","Cross-validation term hits spectral estimator MSE at zero","Bandwidth rule tied to long-run variance accuracy","Local-to-zero CVLL converges to spectral MSE","Proof links local CVLL to zero-frequency error"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000547,"raw_usage":{"total_tokens":2478,"prompt_tokens":801,"completion_tokens":1677,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":545,"completion_tokens_details":{"reasoning_tokens":1606}},"tokens_in":545,"tokens_out":1677,"duration_ms":12206,"temperature":1.0,"reasoning_tokens":1606,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T20:39:42.045638+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A simulation or exact asymptotic computation for a non-Gaussian linear process (for example, an ARMA model with Student-t innovations) that evaluates E[CVLL_c(τ) − L~] and compares it with c0(τ) over τ would decide the matter. If the difference between the two does not vanish relative to n^{4/5−c} for c in (4/5, 1), then the unproved terms are not negligible and the bandwidth selector would be biased. One could also directly compute the expectations of (0.16) and (0.17) at low frequencies; if they are not o(n^{c−4/5}), the theorem's premise fails.","supporting_citations":[],"review_version":1}