Pith. sign in

REVIEW 5 major objections 7 minor 65 references

Robust estimation of the autocorrelation function via forward ratios

T0 review · 5 major / 7 minor · reviewed 2026-07-30 · grok-4.5

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Forward ratios of observations yield simple, outlier-resistant autocorrelation estimates with usable significance bands.

desk verdict Useful robust ACF plug-in with real empirical bite, but the Cauchy/null-variance theory is only airtight under Gaussianity and the higher-lag edge over Ma–Genton is the main incremental claim. read the letter →

arxiv 2607.23744 v1 pith:3OU3CTBS submitted 2026-07-26 econ.EM

A. Montañés , E. Ruiz This is my paper
classification econ.EM
keywords robustestimationautocorrelationfunctionforwardratiostruncatedCauchyoutliersplug-inestimatorsignificancebandsportmanteautests
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Sample autocorrelations collapse toward zero when a series is hit by large additive outliers, so standard correlograms and portmanteau tests become unreliable. This paper builds three estimators from lag-h forward ratios of the series (or of the mean-centered series): a median estimator, a quasi-maximum-likelihood estimator that treats the truncated ratios as truncated-Cauchy, and a closed-form plug-in that inverts the truncated-Cauchy mean via a tanh polynomial. Under zero true autocorrelation the first two estimators are asymptotically normal with known variances, and the plug-in’s variance is nearly the same as the QML variance, so pointwise bands and simple significance tests become available without numerical optimization. Monte Carlo evidence shows the ratio estimators keep bias small under isolated additive and temporary-change outliers where the ordinary sample autocorrelation is badly biased, and they outperform a popular rank-based robust estimator at lags greater than one. Three empirical correlograms (IBEX35 returns, US GDP growth, US inflation) change materially once the plug-in is used, altering conclusions about market efficiency, growth persistence, and inflation memory.

What carries the argument

Forward ratios mapped into [-1,1] and treated as truncated Cauchy with mode equal to ρ(h). The plug-in ˆρ_P(h)=tanh(2.745 z̄ + 1.034 z̄³) inverts the truncated mean without optimization and carries almost the same asymptotic efficiency as the QML estimator under ρ(h)=0.

What would settle it

Simulate stationary ARMA series from a clearly non-elliptical innovation law (for example strongly skewed or multimodal), contaminate them with additive outliers, and check whether the plug-in still stays near the true ρ(h) while ordinary sample autocorrelations collapse and whether the claimed N(0,2.06/T) bands retain correct coverage under ρ(h)=0.

Watch

Extended reading notes

Core claim

The autocorrelation of order h can be recovered from the median, the truncated-Cauchy QML, or a simple plug-in of the mean of the bounded forward ratios z_t = sign(y_t y_{t-h}) min(|y_t|,|y_{t-h}|)/max(|y_t|,|y_{t-h}|). When the true autocorrelation is zero these estimators are asymptotically normal; the plug-in has asymptotic variance approximately 2.06/T, close enough to the QML variance 2/T that ordinary pointwise significance bands can be drawn, while remaining highly resistant to additive outliers.

Load-bearing premise

The series must follow an elliptically contoured law so that lag-h ratios are exactly (truncated) Cauchy with median or mode equal to the true autocorrelation; the proven asymptotics also cover only the zero-autocorrelation case.

Editorial extensions

If this is right

  • Pointwise 95% bands ±1.96√(2/T) (or the plug-in 2.06 variant) can be attached to robust correlograms without relying on fourth-moment Bartlett formulas.
  • Portmanteau statistics built from the plug-in autocorrelations remain usable under outlier contamination where Box–Pierce/Ljung–Box based on sample autocorrelations lose size and power.
  • At lags greater than one the ratio estimators dominate the Ma–Genton rank estimator in finite samples under additive outliers.
  • Empirical conclusions about EMH for equity returns, the strength of GDP-growth dependence, and the persistence of inflation can reverse once outliers are handled by the plug-in rather than by ordinary sample autocorrelations.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the ratios are formed lag by lag, the same construction immediately supplies robust partial-autocorrelation estimates once the Yule–Walker map is applied to the plug-in ACF.
  • If the elliptical-Cauchy step can be relaxed to uncorrelated but conditionally heteroscedastic noise, the method would directly cover daily financial returns without a separate GARCH filter.
  • Joint asymptotic covariance of the vector of plug-in autocorrelations, once derived, would justify a fully robust Ljung–Box-type test rather than the current lag-by-lag bands.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 7 minor

Summary. The paper proposes three robust estimators of the autocorrelation function (ACF) based on forward ratios y_{t+h}/y_t: a median (Hurwicz-type) estimator, a (quasi-)ML estimator built on the truncated-Cauchy law of the sign-adjusted min/max ratio z_t ∈ [−1,1], and a plug-in estimator ρ̂_P(h) = tanh(2.745 z̄ + 1.034 z̄³) that inverts the truncated-Cauchy mean g(ρ) without numerical optimization. Under the null ρ(h)=0 with independent elliptically distributed (Pearson II/VII, Kotz) innovations, the paper derives √T-normal limits with variances π²/4 (median), 2 (ML/QML), and 2.06 (plug-in), yielding pointwise significance bands. Monte Carlo work (T=50–200, AR(1)/MA(1), additive and temporary-change outliers) shows the ratio-based estimators are highly outlier-resistant and that the Ma–Genton rank estimator deteriorates at lag h=5. Three applications (IBEX35 returns, US GDP growth, US inflation) show conclusions change relative to sample autocorrelations. The estimation idea is attractive and the simulations are persuasive on robustness; however, the inference theory has gaps at its foundations, detailed below.

Significance. If the inference theory is repaired, the contribution is useful and practical: a closed-form, optimization-free robust ACF estimator with explicit significance bands would be a genuine addition, since the main competitor (Ma–Genton) has no known asymptotic distribution. The documented degradation of the Ma–Genton estimator at higher lags (Figure 1) is a practically relevant finding, and the three empirical illustrations convincingly show that robust ACF estimation reverses substantive conclusions (EMH for IBEX35, dependence in GDP growth, inflation persistence). The plug-in map has only two fixed, data-independent constants and the delta-method variance calculation is transparent; the simulation design is standard and reproducible in principle, though the Matlab code is not made available. These strengths are real but currently rest on a distributional premise that, as stated, is satisfiable essentially only under Gaussianity (Major Comment 1), which limits the inferential claims exactly in the heavy-tailed settings the method targets.

major comments (5)
  1. [§3, Theorems 3.1–3.3 and Endnote 9] The premise that y_{t+h}/y_t is Cauchy (and z_t truncated Cauchy) invokes Nadarajah (2006), whose result requires the PAIR (y_t, y_{t+h}) to be jointly elliptical. But the theorems simultaneously assume independent innovations (needed for the thinning in Thm 3.2 and the CLT in Thm 3.3). Within an elliptical family, independent components exist only in the Gaussian case, so the assumptions as stated are jointly satisfiable essentially only under Gaussianity. For i.i.d. Student-t7 innovations — the paper's own heavy-tailed design (§4.1) — (y_t, y_{t+h}) is not jointly elliptical and the ratio is not Cauchy; hence Var(z)=(4−π)/π, I(0)=1/2, the 2.06 constant, and the bands (22) and (28) are unproven in the non-Gaussian case. Endnote 9 ('regardless of whether the innovations are independent or just uncorrelated') compounds this: uncorrelatedness does not deliver joint ellipticality either. Th
  2. [§3.2, Theorem 3.2, Eqs. (25)–(28)] The ML estimator (25) is defined on the thinned subsample z*_t with T* = ⌊T/(h+1)⌋+1 observations. Standard ML theory gives √T*(ρ̂−ρ) → N(0, I(0)^{-1}) = N(0,2), hence √T(ρ̂−ρ) → N(0, 2(h+1)), not N(0,2) as stated in (26). The claimed variance 2 can only pertain to the QML estimator (24) that uses all T observations, but for that estimator no proof is given: one must show the lag-h covariance of the scores 2z_t/(1+z_t²) vanishes and derive the sandwich variance. The assertion following the proof that (24) and (25) are 'equivalent when T→∞' is incorrect in efficiency terms — thinning discards a fraction h/(h+1) of the data. As written, the bands ±1.96√(2/T) in (28) are attached to an estimator whose variance is misstated by a factor (h+1).
  3. [§3.1, Lemma 3.1] Median unbiasedness of ρ̂*_H(h) for a general stationary linear process is said to 'follow directly' from Zieliński (1999) and Luger (2006). Those results concern the AR(1) model at h=1, where y_{t+1} − φy_t = ε_{t+1} is independent of y_t. For a general ARMA at lag h, y_{t+h} − ρ(h)y_t is uncorrelated with y_t but not independent of it (e.g., for MA(1) at h=1 it shares ε_t and ε_{t−1} with y_t), so med(y_{t+h}/y_t) = ρ(h) is asserted, not proved. The simulations (Tables 1–2) are consistent with approximate median unbiasedness, but the lemma should either be proved under stated conditions (e.g., a conditional-symmetry requirement) or restricted to the AR class where the argument goes through.
  4. [Table 4, T=100, ϕ=0.3 panel] The H, ML and MR rows in this panel exactly duplicate the corresponding SIZE entries of Table 3 (T=100): e.g., the H row reads 0.093/0.057/0.080/0.031/0.064, identical to Table 3. Power under ϕ=0.3 cannot equal size under ϕ=0 (compare the SAC row, 0.484 vs 0.069). The power comparison for the robust portmanteau tests at T=100, ϕ=0.3 is therefore unreliable as reported; please recompute and correct, and check the remaining panels for similar copy errors.
  5. [§3.1 (after Eq. (23)) and §5] Two inferential extensions are used without adequate support. (i) The claim that the mean-centered estimator ρ̂_H(h) has the same limit law as ρ̂*_H(h) rests on Slutsky/CMT applied to ratios and on the assertion that the median of centered and uncentered ratios is attained at the same time index τ; the ordering of ratios changes under centering and the denominators are heavy-tailed near zero, so a rigorous argument (or explicit simulation evidence on band coverage with estimated mean) is needed. (ii) The empirical sections report p-values for Box–Pierce statistics computed from ρ̂_P (e.g., 7.03, p=0.72 in §5.1) using χ² calibration, although §4.2 acknowledges that the joint asymptotic distribution of (ρ̂_P(1),…,ρ̂_P(H)) is not derived. The conjectural status of these p-values should be stated where they are used, or the joint null covariance should be derived (plausibly diagonal under sy
minor comments (7)
  1. [§3.1] The sentence 'the median of the uncentered ratios in (30)' should refer to Eq. (18); (30) is the plug-in estimator.
  2. [§3.3, proof of Theorem 3.3] The covariance computation writes E(z_t z_{t−h}) = E[(y_{t+h}/y_t)(y_t/y_{t−h})], which substitutes the raw ratios x_t for the truncated z_t; as displayed the identity is not correct. The conclusion Cov(z_t, z_{t−h})=0 does hold under symmetric i.i.d. innovations by a sign-flip argument (conditional on all but y_t, z_t is symmetric about zero), but the proof should be rewritten accordingly.
  3. [Tables 3–4] Notation in the table headers (ML, MR, ˆρ_ML, ˆρ_MR) is inconsistent with the text (QML ˆρ_Q, plug-in ˆρ_P). Also, Table 4 is referenced only implicitly; the size/power discussion in §4.2 would benefit from explicit cross-referencing.
  4. [Figures 4–5] Figure 5's caption says 'Quarterly US inflation' but the series is monthly (m-o-m inflation, Jan 1960–Jun 2026). Please also state sample sizes and the number of lags plotted in each correlogram.
  5. [§4.1] The design states that innovations are generated as standard normal OR standardized Student-t7, but Tables 1–2 and Figures 1–2 report only Gaussian results. The t7 results should be reported — they are directly relevant to Major Comment 1 — including coverage of the asymptotic bands.
  6. [Reproducibility] The paper states results are obtained with 'Matlab codes programmed by the first author'. Posting the code (and the plug-in constant calibration behind a=2.745, b=1.034) in a public repository would materially strengthen the paper.
  7. [Throughout] Typographical issues: 'reallistic', 'lye in the interval', 'seprated', 'recomend', 'probatility', 'extremelly', 'strcutural', 'relatiosnhips', 'Introdcution'; reference list: 'Hannan and Hypde' (Hyde), 'Johson and Kotz' (Johnson), 'Berkoun and Fellaf (2011)' vs 'Berkoun and Fellag', 'Giaritis' (Giraitis), 'Barsky. 1987', 'OECV'; decimal commas in Tables 1–2 ('0,817', '0,204', '-0-272'). The opening sentence of the abstract ('It is obvious to say that…') should be reworded.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: estimators are defined from data ratios, null asymptotics follow from score/information and delta method under stated premises, and plug-in constants are a numerical approximation to a known inverse—not a fit-to-predict loop.

full rationale

The derivation chain is self-contained and non-circular. The three estimators (median of lag-h ratios, QML on truncated-Cauchy z_t, and the plug-in tanh map of z̄) are operational definitions from the sample; they do not define ρ(h) in terms of itself. Asymptotic null variances in Theorems 3.1–3.3 are obtained from the Cauchy/truncated-Cauchy density (via Nadarajah), the binomial indicator variance for the median CLT, the Fisher information integral I(0)=1/2 for the ML score, and a standard delta-method calculation for the plug-in—none of which bake the target ACF into the variance formula. The constants 2.745 and 1.034 in (30) minimize approximation MSE of g^{-1} over ρ; that is ordinary numerical approximation of a known closed-form mean map, not fitting to series data and then calling the fit a prediction. Citations (Hurwicz, Reschenhofer, Zieliński, Luger, Nadarajah, Ma–Genton, Berkoun–Fellag) are external; there is no load-bearing self-citation uniqueness theorem. Correctness risks (joint ellipticity vs independent innovations; median-unbiasedness only proved for AR(1) h=1; asymptotics only under ρ=0) are premise/scope issues, not circular reductions. Empirical illustrations and Monte Carlo comparisons are external checks, not definitional loops. Score 0 with empty steps is the honest finding.

Assumptions & free parameters 2 free parameters · 6 assumptions · 0 invented entities

The contribution is methodological: ratio-based ACF estimators plus null limit laws. It rests on classical stationarity/white-noise structure, the Nadarajah elliptical-ratio→Cauchy link, truncated-Cauchy moments, and two hand-fitted plug-in constants. No new physical entities; free parameters are only the approximation coefficients in the recommended estimator.

free parameters (2)
  • plug-in linear coefficient a=2.745 = 2.745
    Chosen to minimize MSE of the algebraic inverse of g(ρ) over ρ∈(-1,1); enters ˆρ_P and the delta-method variance via g'(0)=a.
  • plug-in cubic coefficient b=1.034 = 1.034
    Same MSE-minimizing fit for tanh(a z̄ + b z̄³); affects finite-sample map though derivative at 0 depends only on a.
assumptions (6)
  • domain assumption Stationary linear process y_t=μ+∑ c_i ε_{t-i} with ∑ c_i²<∞ and white-noise innovations.
    Model (1); defines population ρ(h) the estimators target.
  • domain assumption Innovations (or y_t) elliptically contoured Pearson-II / VII / Kotz-type so lag-h ratios are Cauchy with median ρ(h).
    Invoked via Nadarajah (2006) for density (21) and all of Theorems 3.1–3.3.
  • standard math Zero-median innovations imply median-unbiasedness of the Hurwicz-type estimator (Zieliński/Luger).
    Lemma 3.1 foundation for ˆρ_H.
  • ad hoc to paper QML treats truncated ratios as if independent; ML thins to every (h+1)-st z_t for a proper likelihood under ρ(h)=0.
    §3.2; justifies information calculation I(0)=1/2 and asymptotic variance 2.
  • standard math Sample mean centering does not change the null asymptotic distribution of ratio estimators (Slutsky + continuous mapping).
    Argument after (23) for unknown μ.
  • standard math Truncated standard Cauchy has mean 0 and variance (4-π)/π; uncorrelated z_t when ρ=0.
    Used in Theorem 3.3 proof for √T z̄ and delta method.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Robust estimation of the autocorrelation function via forward ratios." pith.science (2026). https://pith.science/paper/3OU3CTBS

@misc{pith2026260723744,
  author       = {Pith},
  title        = {Pith review of: Robust estimation of the autocorrelation function via forward ratios},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3OU3CTBS}},
  note         = {Machine review of arXiv:2607.23744}
}
read the original abstract

It is obvious to say that an adequate estimation of the autocorrelation function is central in time series analysis. In this paper, we propose three new robust estimators based on ratios of observations, which offer strong resistance against outliers. While the first estimator, which is based on the median, is not efficient, the second is a Quasi Maximum Likelihood (QML) estimator with better efficiency properties. The third estimator is a plug-in estimator, which does not require numerical optimization and, consequently, is extremely simple from a computationally point of view, having similar efficiency to that of the ML estimator. We derive the asymptotic distribution of the first two estimators, when the true autocorrelations are zero. Furthermore, we also show that the asymptotic distribution of the plug-in estimator is rather close to that of the QML estimator, allowing for inference and, in particular, for the construction of point-wise significance bands for the autocorrelations. Using Monte Carlo simulations, we analyse the finite sample properties of the proposed estimators and compare them with those of the sample autocorrelations and alternative extant robust estimators based on ranks. Although the proposed estimators have larger dispersion than the sample autocorrelations in uncontaminated time series, they are highly robust in the presence of outliers. Also, they have better properties than popular alternative robust estimators based on ranks when estimating autocorrelations of order larger than one. The results are illustrated by estimating the correlogram of daily IBEX35 returns, quarterly US economic growth and monthly US inflation.

Figures

Figures reproduced from arXiv: 2607.23744 by the authors.

Figure 1
Figure 1. Monte Carlo densities of estimators of ρ(5) = 0.59 (vertical black lines) obtained when the simulated series of size T = 50 (first column), T = 100 (second column) and T = 300 (third column) are generated with the AR(1) model with ϕ = 0.9 and contaminated with isolated additive outliers with δ = 0 (top row), δ = 3 and λ = 0 (middle row), and δ = 3 and λ = 0.5 (bottom row): r(5) (blue), ˆρMG(5) (orange), ˆρH(5) (yell… view at source ↗
Figure 2
Figure 2. Monte Carlo densities of estimators of ρ(5) = 0 obtained when the simulated series of size T = 50 (first column) and T = 100 (second column) are generated with the AR(1) model with ϕ = 0 and contaminated with isolated additive outliers with δ = 0 (top row), δ = 3 and λ = 0 (middle row), and δ = 3 and λ = 0.5 (bottom row): r(5) (blue), ˆρMG(5) (orange), ˆρH(5) (yellow), ˆρQ(5) (purple), and ˆρP (5) (green). The black… view at source ↗
Figure 3
Figure 3. Daily IBEX35 returns (upper panel), and estimated sample autocorrelations (lower left [PITH_FULL_IMAGE:figures/full_fig_p027_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Quarterly US growth (upper panel), and estimated sample autocorrelations (lower left [PITH_FULL_IMAGE:figures/full_fig_p029_4.png]
Figure 5
Figure 5. Figure 5: Quarterly US inflation (upper panel), and estimated sample autocorrelations (lower left [PITH_FULL_IMAGE:figures/full_fig_p031_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

65 extracted references · 2 linked inside Pith

  1. [1]

    Caiado and N

    Albino, A., J. Caiado and N. Crato (2024), Time series clustering using fragmented autocorrelations,Physica A, 650, 129981

    2024

  2. [2]

    Anderson, T.W. and A.M. Walker (1964), On the asymptotic distribution of the autocorrelations of a sample from a linear stochastic process,Annals of Mathematical Statistics, 35, 1296–1303

    1964

  3. [3]

    (1993), Exactly median-unbiased estimation of first order autore- gressive/unit root models,Econometrica, 61(1), 139–165

    Andrews, D.W.K. (1993), Exactly median-unbiased estimation of first order autore- gressive/unit root models,Econometrica, 61(1), 139–165

    1993

  4. [4]

    (1900),Theorie de la Speculation, Gauthier-Villars, Paris, France

    Bachelier, L. (1900),Theorie de la Speculation, Gauthier-Villars, Paris, France

    1900

  5. [5]

    Cecchetti and R.J

    Ball, L., S.G. Cecchetti and R.J. Gordon (1990), Inflation and uncertainty at short and long horizons,Brooking Papers on Economic Activity, 1, 215–254

    1990

  6. [6]

    Barsky, R,B, (1987), The Fisher hypothesis and the forecastability and persistence of inflation,Journal of Monetary Economics, 19(1), 3–24

    1987

  7. [7]

    (1935), Some aspects of the time correlation problem in regard to test of significance,Journal of the Royal Statistics Society, 98, 536-543

    Bartlett, M.S. (1935), Some aspects of the time correlation problem in regard to test of significance,Journal of the Royal Statistics Society, 98, 536-543

    1935

  8. [8]

    (1946), On the theoretical specification and sampling properties of autocorrelated time-series,Journal of the Royal Statistical Society, Suppl., 8, 27-41

    Bartlett, M.S. (1946), On the theoretical specification and sampling properties of autocorrelated time-series,Journal of the Royal Statistical Society, Suppl., 8, 27-41

    1946

Show all 65 references
  1. [9]

    Berkoun, Y. and H. Fellag (2011), Hurwicz’s estimator of the autoregressive model with non-normal innovations,Applicationes Mathematicae, 38(2), 211-218

    2011

  2. [10]

    Fellag and R

    Berkoun, Y., H. Fellag and R. Zieli´ nski (2003), Robust testing of serial correlation in AR(1) processes in the presence of a single additive outlier,Communications in Statistics-Theory and Methods, 32(8), 1527–1540

    2003

  3. [11]

    Box, G.E. and D.A. Pierce (1970), Distribution of residual autocorrelations in autoregressive-integrated moving average time series models,Journal of the Ameri- can Statistical Association, 65(332), 1509-1526. 33

    1970

  4. [12]

    Caiado, J. and N. Crato (2026), Classification and clustering of time series with data- driven fragmented statistics,International Journal of Data Science and Analytics, forthcoming

    2026

  5. [13]

    (1995a), Time series outliers and spurious autocorrelations,Journal of Applied Statistical Science, 40-51

    Chan, W.-S. (1995a), Time series outliers and spurious autocorrelations,Journal of Applied Statistical Science, 40-51

  6. [14]

    (1995b), Outliers and financial time series modelling: A cautionary note,Mathematics and Computers in Simulation, 39(3-4), 425–430

    Chan, W.-S. (1995b), Outliers and financial time series modelling: A cautionary note,Mathematics and Computers in Simulation, 39(3-4), 425–430

  7. [15]

    Chang, C.C. and D.N. Politis (2016), Robust autocorrelation estimation,Journal of Computational and Graphical Statistics, 25(1), 144-166

    2016

  8. [16]

    Tiao and C

    Chang, I., G.C. Tiao and C. Chen (1988), Estimation of time series parameters in the presence of outliers,Technometrics, 30(2), 193–204

    1988

  9. [17]

    Chen, C. and L.M. Liu (1993), Forecasting time series with outliers,Journal of Forecasting, 12(1), 13-35

    1993

  10. [18]

    Croux, C. and P.J. Rousseeuw (1992), Time-efficient algorithms for two highly robust estimators of scale,Computational Statistics, 2, 411–428

    1992

  11. [19]

    Staneski and N.R

    Dahiya, R.C., P.G. Staneski and N.R. Chaganthy (2001), Maximum likelihood es- timation of parameters of the truncated Cauchy distribution,Communications in Statistics-Theory and Methods, 30(8-9), 1737–1750

    2001

  12. [20]

    Giraitis and P.C.B

    Dalla, V., L. Giraitis and P.C.B. Phillips (2022), Robust tests for white noise and cross-correlation,Econometric Theory, 38, 913–941

    2022

  13. [21]

    Escribano and P

    del Barrio Castro, T., A. Escribano and P. Sibbertsen (2025), Modeling and forecast- ing the long memory of cyclical transition in paleoclimate data,Energy Economics, 147, 108520

    2025

  14. [22]

    Diebold, F.X. and A. Inoue (2001), Long memory and regime switching,Journal of Econometrics, 105, 131-159

    2001

  15. [23]

    (1970), Testing for serial correlation in least-squares regression when some of the regressors are lagged dependent variables,Econometrics, 38, 410–421

    Durbin, J. (1970), Testing for serial correlation in least-squares regression when some of the regressors are lagged dependent variables,Econometrics, 38, 410–421. 34

    1970

  16. [24]

    Fried and T

    D¨ urre, A., R. Fried and T. Liboschik (2015), Robust estimation of (partial) autocor- relation,Wiley Interdisciplinary Reviews: Computational Statistics, 7(3), 205–222

    2015

  17. [25]

    (1965), The behaviour of stock market prices,Journal of Business, 38, 34-105

    Fama, E.F. (1965), The behaviour of stock market prices,Journal of Business, 38, 34-105

    1965

  18. [26]

    Fellag, H. and R. Zieli´ nski (1996), Bias of the LSE estimator of the first order au- toregressive model under Tukey contamination,Communications in Statistics-Theory and Methods, 25(7), 1537–1551

    1996

  19. [27]

    Franses, P.H. and N. Haldrup (1994), The effects of additive outliers on tests for unit roots and cointegration,Journal of Business & Economic Statistics, 12(4), 471-478

    1994

  20. [28]

    Fuhrer, J,C, (2010), Inflation persistence, in Friedman, B.M. and M. Woodford (eds.), Handbook of Monetary Economics, vol. 3, 423–486

    2010

  21. [29]

    Gadea, M.D. and L. Mayoral (2006), The persistence of inflation in OECV countries: A fractionally integrated approach,Journal of Central Banking, 2(1), 51–104

    2006

  22. [30]

    Golinelli, R. and G. Parigi (2007), The use of monthly indicators to forecast quarterly GDP in the short run: An application to G7 countries,Journal of Forecasting, 26, 77–94

    2007

  23. [31]

    Lopez Gaffney and S

    Gospodinov, N., I. Lopez Gaffney and S. Ng (2025), The economic impact of low- and high-frequency temperature changes, arXiv:2505.08950v1[econ.GN]

    2025 arXiv

  24. [32]

    and N, Hyung (1999), Occasional strcutural breaks and long mem- ory, Discussion paper no

    Granger, C.W.J. and N, Hyung (1999), Occasional strcutural breaks and long mem- ory, Discussion paper no. 99-14, University of California, San Diego

    1999

  25. [33]

    (2000), Robust estimation for the coefficients of a first order autoregressive process,Communications in Statistics-Theory and Methods, 29(1), 45–54

    Guo, J.-H. (2000), Robust estimation for the coefficients of a first order autoregressive process,Communications in Statistics-Theory and Methods, 29(1), 45–54

    2000

  26. [34]

    Guo, J.-H. and L. Billard (2012), Assessing one-step-ahead prediction error based on the median of first-order autoregressive models in the presence of outliers,Commu- nications in Statistics-Theory and Methods, 41(15), 2738–2749. 35

    2012

  27. [35]

    Guttman, I. and G.C. Tiao (1978), Effect of correlation on the estimation of a mean in the presence of spurious observations,The Canadian Journal of Statistics, 6(2), 229–247

    1978

  28. [36]

    (2000), On robust estimation in the first-order autoregressive process, Communications in Statistics-Theory and Methods, 29(1), 45–54

    Haddad, J.N. (2000), On robust estimation in the first-order autoregressive process, Communications in Statistics-Theory and Methods, 29(1), 45–54

    2000

  29. [37]

    Hannan, E.J. and C.C. Hypde (1972), On limit theorems for quadratic functions of discrete time series,The Annals of Mathematical Statistics, 43, 2058–2066

    1972

  30. [38]

    Hassler, U. and J. Wolters (1995), Long-memory in inflation rates: International evidence,Journal of Business & Economic Statistics, 13(1), 37–45

    1995

  31. [39]

    Pohle and T

    Hassler, U., M.-O. Pohle and T. Zahn (2025), Simultaneous inference bands for autocorrelations, arXiv:2503.18560v2[econ.EM]

    2025 arXiv

  32. [40]

    (1950), Least-squares bias in time series, in Koopmans, T.C

    Hurwicz, L. (1950), Least-squares bias in time series, in Koopmans, T.C. (ed.),Sta- tistical Inference in Dynamic Economic Models, Wiley, New York

    1950

  33. [41]

    Johson, N.L. and S. Kotz (1970),Continuous Univariate Distributions, vol. I, Wiley- and Sons, New York

    1970

  34. [42]

    Kwan, A.C.C. and A. Sim (1996), On the finite-sample distribution of modified portmanteau tests for randomness of a Gaussian time series,Biometrika, 83(4), 938- 946

    1996

  35. [43]

    Ljung, G.M. and G.E. Box (1978), On a measure of lack of fit in time series models, Biometrika, 65(2), 297-303

    1978

  36. [44]

    Nankervis and N.E

    Lobato, I., J.L. Nankervis and N.E. Savin (2001), Testing for autocorrelation using a modified Box-Pierce Q test.International Economic Review, 42(1), 187-205

    2001

  37. [45]

    Luger, R. (2006), Median-unbiased estimation and exact inference methods for first- order autoregressive models with conditional heteroscedasticity of unknown form, Journal of Time Series Analysis, 27(1), 119–128

    2006

  38. [46]

    Ma, Y. and M. Genton (2000), Highly robust estimation of the autocovariance func- tion,Journal of Time Series Analysis, 21, 663-684. 36

    2000

  39. [47]

    Mann, H.B. and A. Wald (1943), On stochastic limit and order relatiosnhips,Annals of Mathematical Statistics, 14(3), 217-226

    1943

  40. [48]

    Mikosh, T. and C. Starica (2004), Nonstationarity in financial time series, the long- run dependence and the IGARCH model,The Review of Economics and Statistics, 86(1), 378-390

    2004

  41. [49]

    (1980), Comment on ”Robust estimation of autoregressive models”, in Brillinger, D.R

    Miller, R.B. (1980), Comment on ”Robust estimation of autoregressive models”, in Brillinger, D.R. and G.C. Tiao (eds.),Direction in Time Series, Institute of Mathe- matical Statistics, Hayward, CA

    1980

  42. [50]

    (2006), On the ratioX/Yfor some elliptically symmetric distributions, Journal of Multivariate Analysis, 97(2), 342-358

    Nadarajah, S. (2006), On the ratioX/Yfor some elliptically symmetric distributions, Journal of Multivariate Analysis, 97(2), 342-358

    2006

  43. [51]

    Nadarajah, S. and S. Kotz (2006), A Truncated Cauchy Distribution,International Journal of Mathematical Education in Science & Technology, 37(5), 605-608

    2006

  44. [52]

    (2006), Correlograms for non-stationary autoregressions,Journal of the Royal Statistical Society, Series B, 68, 707–720

    Nielsen, B. (2006), Correlograms for non-stationary autoregressions,Journal of the Royal Statistical Society, Series B, 68, 707–720

    2006

  45. [53]

    Phillips, P.C. and V. Solo (1992), Asymptotics for linear processes,Annals of Statis- tics, 20, 971–1001

    1992

  46. [54]

    Pivetta, F. and R. Reiss (2007), The persistence of inflation in the United States, Journal of Economic Dynamics and Control, 31(4), 1326–1358

    2007

  47. [55]

    Proietti, T. and A. Luati (2026), Separating long- memory from short, Manuscript

    2026

  48. [56]

    (2019), Heteroscedasticity-robust estimation of autocorrelation, Communications in Statistics-Simulation and Computation, 48(4), 1251–1263

    Reschenhofer, E. (2019), Heteroscedasticity-robust estimation of autocorrelation, Communications in Statistics-Simulation and Computation, 48(4), 1251–1263

    2019

  49. [57]

    Reschenhofer, E. and M.A. Hauser (1997), Tests of the Efficient Markets Hypothesis, Austrian Journal of Statistics, 26(1), 31-52

    1997

  50. [58]

    (1976),An Introdcution to Probability Theory and Mathematical Statistics, Wiley and Sons, New York

    Rohatgi, V.K. (1976),An Introdcution to Probability Theory and Mathematical Statistics, Wiley and Sons, New York

    1976

  51. [59]

    Rousseeuw, P.J. and C. Croux (1993), Alternatives to the median absolute deviation, Journal of the American Statistical Association, 88(424), 1273–1283. 37

    1993

  52. [60]

    Jaber and A.G

    Smadi, A.A., J.J. Jaber and A.G. AlZubi (2014), Robustness of several estimators of the acf of AR(1) process with non-Gaussian errors,Journal of Modern Applied Statistical Methods, 13(1), article 10

    2014

  53. [61]

    (2018), Hurwicz estimator for autoregressive model with gen- eralized error distributed innovations,Journal of the Indian Society for Probability and Statistics, 19, 299-320

    Sri Ranganath, C.G. (2018), Hurwicz estimator for autoregressive model with gen- eralized error distributed innovations,Journal of the Indian Society for Probability and Statistics, 19, 299-320

    2018

  54. [62]

    (1986), Time series model specification in the presence of outliers,Journal of the American Statistical Association, 81(393), 132–141

    Tsay, R.S. (1986), Time series model specification in the presence of outliers,Journal of the American Statistical Association, 81(393), 132–141

    1986

  55. [63]

    Vogelsang, T.J. and J. Wang (2016), Exactly/nearly unbiased estimation of auto- covariances of a univariate time series with unknown mean,Journal of Time Series Analysis, 37(6), 723–740

    2016

  56. [64]

    (1921), On the time-correlation problem with especial reference to the variate-difference correlation method,Journal of the Royal Statistical Society, 84, 497–537

    Yule, G.U. (1921), On the time-correlation problem with especial reference to the variate-difference correlation method,Journal of the Royal Statistical Society, 84, 497–537

    1921

  57. [65]

    (1999), A median-unbiased estimator of the AR(1) coefficient,Journal of Time Series Analysis, 20, 477–481

    Zieli´ nski, R. (1999), A median-unbiased estimator of the AR(1) coefficient,Journal of Time Series Analysis, 20, 477–481. Notes 1Nielsen (2006), shows thatr(h) could deliver misleading inferences in non-stationary frameworks. 2The classical restrictions for the asymptotic dis...

    1999

Pith tools

Reviewed July 30, 2026 · model on record in the stance chip above.