Estimation of the Risk Measure under a Nuisance Autoregression

Jana Jure\v{c}kov\'a, Jan Picek

Pith reviewed 2026-05-12 04:39 UTC · model grok-4.3

classification 📊 stat.ME

keywords autoregressionR-estimatorsquantile functionsrisk measurestime serieserror term estimation

0 comments

The pith

Estimates of quantile functions of the unobservable error term Z are constructed from autoregressive observations using R-estimators.

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

The paper develops a method to estimate various quantile functions of the error term Z_t in an autoregressive model, where Z_t is the unobservable increment of the observed series X_t beyond its past values. Since only the X series is available, the authors use R-estimators of the unknown autoregression coefficients together with autoregression quantiles to recover the quantiles of Z. This setup arises in experiments that track profit, loss, or physical quantities over time, where current measurements depend on previous ones. A reader would care because it enables risk assessment and economic indicators based on the pure increments without direct observation of Z.

Core claim

We construct an estimate of quantile functions of Z in the situation that the inference is possible only by means of observations X. The proposed estimates are based on the R-estimators of autoregression coefficients, combined with the autoregression quantiles.

What carries the argument

R-estimators of autoregression coefficients combined with autoregression quantiles to recover quantile functions of the unobservable error term Z_t.

Load-bearing premise

The autoregressive model is correctly specified, the R-estimators are consistent for the coefficients, and the autoregression quantiles can be validly combined to recover the quantiles of Z.

What would settle it

Simulate an autoregressive process with a known error distribution, apply the proposed estimators to the generated X series, and check whether the recovered quantiles match the known quantiles of Z within sampling variability.

read the original abstract

The goal of an experiment is to evaluate the profit, loss, or the amount of a physical entity over a period. The measurements $X_t$ can be influenced by the values measured in the past; hence we describe the situation with an autoregression model, whose autoregression coefficients are generally unknown. The variable of interest is the error term $Z_t$ of the model, which is the increment of $X_t$ with respect to the past, but itself unobservable. The problem is to estimate various quantile functions of $Z$, as the risk measure of the loss or the related economic indicators. We construct an estimate of quantile functions of $Z$ in the situation that the inference is possible only by means of observations $X$. The proposed estimates are based on the R-estimators of autoregression coefficients, combined with the autoregression quantiles.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This paper constructs quantile estimates for the unobservable AR innovations by plugging R-estimates of the coefficients into autoregression quantile methods.

read the letter

The main contribution is a direct construction for recovering quantile functions of Z_t, the innovation in an autoregressive model, when only the observed series X_t is available. They first estimate the AR coefficients with R-estimators and then feed those into the autoregression quantile framework to target the quantiles of Z. This targets a concrete problem in risk measurement where the increments matter but are not directly seen. R-estimators bring robustness to the coefficient step, which is a reasonable choice for time series that may have outliers or heavy tails. The overall setup is clean and stays within standard tools for dependent data. The paper does a service by spelling out how the two pieces fit together for this risk-measure application. The assumptions are the usual ones: correct AR specification and consistency of the R-estimators, which are standard in this literature. The central claim holds up on its own terms as a methodological recipe rather than a deep theoretical advance. The soft spots are modest. The abstract gives no derivation or asymptotic result, so the propagation of coefficient estimation error into the quantile estimates is not visible here; if the full paper lacks explicit rates or simulation checks, that would leave the practical performance unclear. Novelty looks incremental rather than transformative, since it recombines existing R-estimator and quantile autoregression ideas. No circularity or invented entities appear. This is for statisticians working on robust time-series methods or applied researchers in finance and risk who need quantile tools for increments. A reader already familiar with quantile autoregression will see the extension quickly and can judge whether the combination adds enough for their setting. It deserves a serious referee because the problem is well-posed, the components are credible, and referees can verify the technical steps and any supporting evidence.

Referee Report

1 major / 0 minor

Summary. The manuscript proposes a construction for estimating quantile functions of the unobservable innovation process Z_t in an autoregressive model X_t = sum phi_i X_{t-i} + Z_t (with unknown coefficients), using only observations of X. The estimators combine R-estimators of the autoregression coefficients with autoregression quantiles to recover risk measures such as quantiles of Z.

Significance. If the estimators are shown to be consistent under standard conditions on the AR model and the R-estimators, the result would extend quantile-based risk measurement to settings with nuisance autoregressive dependence, which is relevant for time-series applications in econometrics and risk analysis.

major comments (1)

[Abstract] Abstract: the central claim asserts that the proposed estimates exist and are based on R-estimators combined with autoregression quantiles, but supplies neither the explicit form of the estimator, the regularity conditions required for consistency, nor any derivation or asymptotic result. This absence prevents evaluation of whether the construction recovers the quantiles of Z_t.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their careful review and for highlighting the need for greater clarity in the abstract. We address the major comment below and will revise the manuscript accordingly.

read point-by-point responses

Referee: [Abstract] Abstract: the central claim asserts that the proposed estimates exist and are based on R-estimators combined with autoregression quantiles, but supplies neither the explicit form of the estimator, the regularity conditions required for consistency, nor any derivation or asymptotic result. This absence prevents evaluation of whether the construction recovers the quantiles of Z_t.

Authors: We agree that the abstract is concise and omits the explicit estimator form, regularity conditions, and asymptotic results, which limits immediate evaluation. The body of the manuscript defines the estimator explicitly as the composition of an R-estimator for the AR coefficients with the autoregression quantile process applied to the residuals, derives consistency under standard conditions (stationary AR process with absolutely summable coefficients, continuous innovation density, and standard regularity on the R-estimator score function), and proves the result via uniform convergence arguments. In the revision we will expand the abstract to state the estimator form in one sentence, list the key regularity conditions, and reference the consistency theorem, thereby allowing readers to assess the recovery of the quantiles of Z_t directly from the abstract. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper constructs estimates of quantile functions of the unobservable error term Z_t from observations X_t in an autoregressive model by combining R-estimators of the unknown autoregression coefficients with autoregression quantiles. This is a standard two-step statistical procedure resting on consistency of R-estimators and validity of combining autoregression quantiles under correct model specification. No step in the described chain reduces by definition, by renaming a fitted quantity as a prediction, or by load-bearing self-citation to the paper's own inputs. The derivation remains self-contained against external benchmarks and standard assumptions.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The abstract supplies no explicit free parameters, axioms, or invented entities; the approach appears to rest on standard statistical concepts whose precise invocation is not detailed here.

pith-pipeline@v0.9.0 · 5448 in / 1162 out tokens · 32766 ms · 2026-05-12T04:39:17.499349+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

22 extracted references · 22 canonical work pages

[1]

Journal Financial Economics 2(4): 477–492

Bassett GW, Jr, Koenker R, Kordas W (2004) Pessimistic portfolio allocation and Choquet Expected Utility. Journal Financial Economics 2(4): 477–492

work page 2004
[2]

Statistics and Public Policy 3(1): 1–19

Gastwirth JL (2016) Measures of Economic Inequality Focusing on the Status of the Lower and Middle Income Groups. Statistics and Public Policy 3(1): 1–19

work page 2016
[3]

Research in Statistics 3(1)

Gastwirth JL, Zhao X (2025) Additional insights provided by alternative measures of eco- nomic inequality: Income inequality in the United States and wealth inequality in the United Kingdom increased more in recent decades than indicated by the Gini coefficient. Research in Statistics 3(1)

work page 2025
[4]

Gutenbrunner C, Jure ˇckov´a J (1992) Regression rank scores and regression quantiles. Ann. Statist. 20(1): 305–330

work page 1992
[5]

(2020) Averaged Autoregression Quantiles in Autoregressive Model

G ¨uney Y , Jureˇckov´a J, Arslan, O. (2020) Averaged Autoregression Quantiles in Autoregressive Model. In: Analytical Methods in Statistics Eds.: M. Maciak et al. Springer Proceedings in Mathematics and Statistics 329, page 115. ISBN 978-3-030-48813-0

work page 2020
[6]

Ann Statist 27(4), 1385–1414

Hallin M, Jure ˇckov´a J (1999) Optimal tests for autoregressive models based on autoregression rank scores. Ann Statist 27(4), 1385–1414

work page 1999
[7]

Doksum: 335–362

Hallin M, Jure ˇckov´a J, Koul HL (2007) Serial Autoregression and Regression Rank Scores Statistics In: Advances In Statistical Modeling and Inference: Essays in Honor of Kjell A. Doksum: 335–362

work page 2007
[8]

Ann Math Statist 43, 1449–1459

Jaeckel LA (1972) Estimating regression coefficients by minimizing the dispersion of the residuals. Ann Math Statist 43, 1449–1459

work page 1972
[9]

Jure ˇckov´a J, Arslan O, G¨uney Y , Picek J, Schindler M, Tuac ¸ Y (2022) Nonparametric tests in linear model with autoregressive errors Metrika 86: 443–453

work page 2022
[10]

arXiv:2212.12419v1

Jure ˇckov´a J, Kalina J, Ve ˇceˇr J (2022) Estimation of expected shortfall under various experi- mental conditions. arXiv:2212.12419v1

work page arXiv 2022
[11]

Jure ˇckov´a J, Koul HL, Picek J (2025) R-estimation in Linear Model with Autoregressive Er- rors.Submitted for publication

work page 2025
[12]

Sankhya 67(2): 227–252

Jure ˇckov´a J, Picek J (2005) Two-step regression quantiles. Sankhya 67(2): 227–252

work page 2005
[13]

Jure ˇckov´a J, Picek J, Kalina J (2024) Estimation of Conditional Value-at-Risk in Linear Model In: Combining, Modelling and Analyzing Imprecision, Randomness and Dependence. J. Ansari et al. (Eds.), AISC [Advances in Intelligent Systems and Computing] 1458, pp. 200– 207, Springer. Estimation of the Risk Measure under a Nuisance Autoregression 11

work page 2024
[14]

Econometrica 46: 33–50

Koenker RJ, Bassett G (1978) Regression quantiles. Econometrica 46: 33–50

work page 1978
[15]

Lecture Notes-Monograph Series 21 IMS Hayward California

Koul HL (1992) Weighted Empiricals and Linear Models. Lecture Notes-Monograph Series 21 IMS Hayward California

work page 1992
[16]

Ann Statist 22: 540–562

Koul HL, Ossiander M (1994) Weak convergence of randomly weighted dependent residual empiricals with applications to autoregression. Ann Statist 22: 540–562

work page 1994
[17]

Ann Statist 23(2): 670–689

Koul HL, Saleh AK (1995) Autoregression quantiles and related rank-scores processes. Ann Statist 23(2): 670–689

work page 1995
[18]

Journal of Multivariate Analysis 81: 167–186

Mukherjee K, Bai ZD (2002) R-estimation in autoregression with square-integrable score function. Journal of Multivariate Analysis 81: 167–186

work page 2002
[19]

The Ameri- can Statistician 72(4): 328–343

Prendergast LA, Staudte RG (2018) A Simple and Effective Inequality Measure. The Ameri- can Statistician 72(4): 328–343

work page 2018
[20]

Journal of Banking & Finance 31: 3524–3538

Trindade AA, Uryasev SP, Shapiro A, Zrazhevsky G (2007) Financial prediction with con- strained tail risk. Journal of Banking & Finance 31: 3524–3538

work page 2007
[21]

Hydrological Processes 16: 1807–1829

Yue S, Pilon P, Phinney B, Cavadias G (2002) The influence of autocorrelation on the ability to detect trend in hydrological series. Hydrological Processes 16: 1807–1829

work page 2002
[22]

Yue S, Pilon P (2003) Interaction between deterministic trend and autoregressive pro- cess.Water Resources Research 39(4): 1077

work page 2003