pith. sign in

arxiv: 2607.00219 · v1 · pith:HAOLXR76new · submitted 2026-06-30 · 💰 econ.EM

Asymptotic Properties of Empirical Quantile-Based Estimators

Julien Chhor , Xavier D'Haultf{\oe}uille , J\'er\'emy L'Hour , Martin Mugnier This is my paper

Pith reviewed 2026-07-02 00:26 UTC · model grok-4.3

classification 💰 econ.EM
keywords quantile estimatorsplug-in estimationasymptotic normalitychanges-in-changesempirical processesroot-n consistencyvariance estimationunbounded support
0
0 comments X

The pith

Plug-in estimator for expected quantile compositions is root-n consistent and asymptotically normal even when variables are unbounded.

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

The paper studies parameters of the form the expected value of the quantile function of Y composed with the distribution function of Z and evaluated at X, which appear in changes-in-changes models. It shows that the natural plug-in estimator using empirical distribution functions is root-n consistent and asymptotically normal under regularity conditions weaker than those in earlier work. The paper also constructs a new estimator of the asymptotic variance of this plug-in estimator and proves its consistency under the same weaker conditions that permit unbounded support. Monte Carlo evidence indicates that the root-n and normality conditions are close to minimal and that the new variance estimator yields more accurate inference than previous alternatives.

Core claim

The plug-in estimator θ̂ of θ₀ = E[F_Y^{-1} ∘ F_Z(X)] is root-n consistent and asymptotically normal under weaker conditions on the distributions and joint law than previously available, and a new consistent estimator of its asymptotic variance is available under the same conditions.

What carries the argument

The plug-in estimator formed by replacing F_Y, F_Z with their empirical counterparts in the expression for θ₀, together with the new consistent estimator of its asymptotic variance.

If this is right

  • Root-n inference becomes available for changes-in-changes parameters even when outcome variables have unbounded support.
  • A consistent variance estimator exists that does not require the stronger boundedness conditions used in prior work.
  • Simulation evidence suggests the stated regularity conditions cannot be relaxed much further while preserving root-n normality.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same plug-in construction and variance estimator may apply to other parameters defined via compositions of quantile and distribution functions.
  • Empirical researchers using changes-in-changes designs with heavy-tailed data can now obtain valid confidence intervals without trimming or truncation.
  • The approach may extend to semiparametric models that combine quantile-based functionals with additional covariates.

Load-bearing premise

The distributions of Y and Z and the joint distribution of (X, Y, Z) satisfy regularity conditions sufficient for the functional delta method or empirical-process arguments to deliver root-n normality.

What would settle it

A Monte Carlo design or empirical example with unbounded variables in which either the plug-in estimator fails to be asymptotically normal or the proposed variance estimator fails to be consistent.

Figures

Figures reproduced from arXiv: 2607.00219 by J\'er\'emy L'Hour, Julien Chhor, Martin Mugnier, Xavier D'Haultf{\oe}uille.

Figure 1
Figure 1. Figure 1: Log of IQR( bθ) as a function of log(N) for various (b2, d2). Next, we investigate in [PITH_FULL_IMAGE:figures/full_fig_p014_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Empirical distribution of √ N( bθ − θ0)/σ (solid) v.s. standard nor￾mal density (dotted). We now turn to inference. We consider six confidence intervals. The first five are based on asymp￾totic normality and different variance estimators, whereas the last relies on the bootstrap distri￾bution. The first variance estimator (Split column) is σb 2 , which is consistent for σ 2 under the assumptions of Theorem… view at source ↗
read the original abstract

We consider inference for parameters of the form $\theta_0 = E[F_Y^{-1}\circ F_Z(X)]$ for some variables $X$, $Y$ and $Z$. Such parameters appear, in particular, in the ``changes-in-changes'' model of \cite{AtheyImbens2006}. We first establish that $\widehat{\theta}$, a plug-in estimator of $\theta_0$, is root-$n$ consistent and asymptotically normal under weaker conditions than those previously available, allowing in particular for unbounded variables. Next, we propose a new estimator of the asymptotic variance of $\widehat{\theta}$ and show its consistency, also allowing for unbounded variables. Monte Carlo simulations suggest that the conditions for root-$n$ consistency and asymptotic normality are, in some sense, minimal. These simulations highlight that our variance estimator also leads to more accurate inference than some alternative approaches.

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.

Referee Report

2 major / 2 minor

Summary. The paper studies the parameter θ₀ = E[F_Y^{-1} ∘ F_Z(X)] arising in models such as changes-in-changes. It establishes root-n consistency and asymptotic normality of the plug-in estimator θ̂ under regularity conditions on the marginal and joint distributions that are weaker than those in prior work and explicitly permit unbounded support. It further proposes a new estimator of the asymptotic variance of θ̂ and proves its consistency under the same conditions. Monte Carlo experiments are used to suggest that the regularity conditions are close to minimal and that the new variance estimator yields more accurate inference than existing alternatives.

Significance. If the derivations are correct, the results would be useful for empirical work with quantile-based functionals that involve unbounded variables, extending the scope of valid inference beyond what is currently available. The provision of a consistent variance estimator and the Monte Carlo evidence on minimality of conditions are practical strengths.

major comments (2)
  1. [§3] §3 (asymptotic normality theorem): the precise statement of the regularity conditions on F_Y, F_Z and the joint law of (X,Y,Z) that replace the bounded-support assumptions of earlier work is not compared side-by-side with the conditions in Athey and Imbens (2006); without this comparison it is difficult to confirm that the new conditions are strictly weaker while still sufficient for the functional delta method.
  2. [§4] §4 (variance estimator): the proof that the proposed variance estimator is consistent relies on uniform convergence arguments that must hold when the support is unbounded; the paper should state explicitly which empirical-process maximal inequalities are invoked and verify that the envelope conditions remain satisfied under the stated moment restrictions.
minor comments (2)
  1. [§5] The abstract claims that Monte Carlo evidence indicates the conditions are 'minimal in some sense'; the simulation design in §5 should be described more precisely (e.g., tail indices used to probe unboundedness) so that readers can assess how close to the boundary the designs actually are.
  2. [§2] Notation for the empirical distribution functions and their inverses should be introduced once in §2 and used consistently thereafter; occasional switches between F̂ and F_n create unnecessary ambiguity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the positive evaluation and recommendation of minor revision. We address each major comment below and will incorporate the suggested clarifications into the revised manuscript.

read point-by-point responses

  1. Referee: [§3] §3 (asymptotic normality theorem): the precise statement of the regularity conditions on F_Y, F_Z and the joint law of (X,Y,Z) that replace the bounded-support assumptions of earlier work is not compared side-by-side with the conditions in Athey and Imbens (2006); without this comparison it is difficult to confirm that the new conditions are strictly weaker while still sufficient for the functional delta method.

    Authors: We agree that an explicit side-by-side comparison would improve readability and verifiability. In the revised manuscript we will insert a new table (or subsection) in §3 that juxtaposes the key regularity conditions from Athey and Imbens (2006) with our own, highlighting the relaxations (in particular the removal of bounded-support requirements) while confirming that the functional delta method still applies under the stated moment and smoothness assumptions. revision: yes

  2. Referee: [§4] §4 (variance estimator): the proof that the proposed variance estimator is consistent relies on uniform convergence arguments that must hold when the support is unbounded; the paper should state explicitly which empirical-process maximal inequalities are invoked and verify that the envelope conditions remain satisfied under the stated moment restrictions.

    Authors: We will revise the proof of consistency for the variance estimator in §4 to name the specific empirical-process maximal inequalities employed (e.g., those in van der Vaart and Wellner, 1996) and to include an explicit verification that the relevant envelope functions satisfy the required integrability conditions under our moment restrictions, even when supports are unbounded. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation self-contained

full rationale

The paper establishes root-n consistency, asymptotic normality, and consistency of a variance estimator for the plug-in estimator of θ₀ = E[F_Y^{-1} ∘ F_Z(X)] by applying the functional delta method and empirical process arguments under regularity conditions on the distributions. These steps rely on standard external theory rather than reducing any prediction or variance estimator to a fitted quantity by construction, and no load-bearing self-citation or ansatz is invoked. The Monte Carlo evidence is presented as supporting minimality of conditions but does not substitute for the analytic derivation. The central claims remain independent of the paper's own inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Review performed on abstract only; no explicit free parameters, axioms, or invented entities are identifiable. Standard empirical process theory and functional delta method are implicitly used but not detailed.

pith-pipeline@v0.9.1-grok · 5695 in / 1001 out tokens · 18184 ms · 2026-07-02T00:26:31.160527+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

37 extracted references · 10 canonical work pages

  1. [1]

    Bulletin of Mathematical Biology , year =

    Arratia, Richard and Gordon, Louis , title =. Bulletin of Mathematical Biology , year =. doi:10.1007/BF02458840 , url =

    work page doi:10.1007/bf02458840

  2. [2]

    2007 , author =

    Measure theory. 2007 , author =. doi:10.1007/978-3-540-34514-5 , isbn =

    work page doi:10.1007/978-3-540-34514-5 2007

  3. [3]

    1991 , url=

    Accounting for the Slowdown in Black-White Wage Convergence , author=. 1991 , url=

    1991

  4. [4]

    2025 , eprint=

    On a Debiased and Semiparametric Efficient Changes-in-Changes Estimator , author=. 2025 , eprint=

    2025

  5. [5]

    and Piketty, Thomas and Saez, Emmanuel , Title =

    Atkinson, Anthony B. and Piketty, Thomas and Saez, Emmanuel , Title =. Journal of Economic Literature , Volume =. 2011 , Month =. doi:10.1257/jel.49.1.3 , URL =

    work page doi:10.1257/jel.49.1.3 2011

  6. [6]

    D. G. Champernowne , journal =. The Graduation of Income Distributions , urldate =

  7. [7]

    Yntema , title =

    Dwight B. Yntema , title =. Journal of the American Statistical Association , volume =. 1933 , publisher =. doi:10.1080/01621459.1933.10503242 , URL =

    work page doi:10.1080/01621459.1933.10503242 1933

  8. [8]

    2009 , journal =

    Power Laws in Economics and Finance , author =. 2009 , journal =

    2009

  9. [9]

    Handbook of Labor Economics , editor =

    Race and gender in the labor market , author =. Handbook of Labor Economics , editor =. 1999 , chapter =

    1999

  10. [10]

    Schennach , keywords =

    Arthur Lewbel and Susanne M. Schennach , keywords =. A simple ordered data estimator for inverse density weighted expectations , journal =. 2007 , issn =. doi:https://doi.org/10.1016/j.jeconom.2005.08.005 , url =

    work page doi:10.1016/j.jeconom.2005.08.005 2007

  11. [11]

    , title =

    Athey, Susan and Imbens, Guido W. , title =. Econometrica , volume =. doi:10.1111/j.1468-0262.2006.00668.x , year =

    work page doi:10.1111/j.1468-0262.2006.00668.x 2006

  12. [12]

    Statistics & Probability Letters , volume=

    A sharp estimate of the binomial mean absolute deviation with applications , author=. Statistics & Probability Letters , volume=. 2013 , publisher=

    2013

  13. [13]

    Probability and measure , url =

    Billingsley,. Probability and measure , url =

  14. [14]

    Memoirs of the American Mathematical Society , year=

    One-dimensional empirical measures, order statistics, and Kantorovich transport distances , author=. Memoirs of the American Mathematical Society , year=

  15. [15]

    2013 , publisher=

    Concentration Inequalities: A Nonasymptotic Theory of Independence , author=. 2013 , publisher=

    2013

  16. [16]

    Casella, George and Berger, Roger , biburl =

  17. [17]

    The Annals of Probability , pages=

    Weighted empirical and quantile processes , author=. The Annals of Probability , pages=

  18. [18]

    2004 , publisher=

    Order Statistics , author=. 2004 , publisher=

    2004

  19. [19]

    arXiv preprint arXiv:2507.07228 , year=

    On a Debiased and Semiparametric Efficient Changes-in-Changes Estimator , author=. arXiv preprint arXiv:2507.07228 , year=

    work page arXiv

  20. [20]

    Fuzzy Differences-in-Differences , journal =

    de Chaisemartin, C and D’Haultf. Fuzzy Differences-in-Differences , journal =. 2018 , month =

    2018

  21. [21]

    The Annals of Probability , pages=

    Necessary and sufficient conditions for asymptotic normality of L-statistics , author=. The Annals of Probability , pages=. 1992 , publisher=

    1992

  22. [22]

    Journal of Business & Economic Statistics , volume=

    A comparison of two quantile models with endogeneity , author=. Journal of Business & Economic Statistics , volume=. 2020 , publisher=

    2020

  23. [23]

    Quantitative Economics , volume=

    Counterfactual mapping and individual treatment effects in nonseparable models with binary endogeneity , author=. Quantitative Economics , volume=. 2017 , publisher=

    2017

  24. [24]

    On the substitution rule for Lebesgue–Stieltjes integrals

    Neil Falkner and Gerald Teschl. On the substitution rule for Lebesgue–Stieltjes integrals. Expositiones Mathematicae. 2012. doi:https://doi.org/10.1016/j.exmath.2012.09.002

    work page doi:10.1016/j.exmath.2012.09.002 2012

  25. [25]

    A Characterization of the Asymptotic Normality of Linear Combinations of Order Statistics from the Uniform Distribution

    Hecker, H. A Characterization of the Asymptotic Normality of Linear Combinations of Order Statistics from the Uniform Distribution. Ann. Statist. 1976. doi:10.1214/aos/1176343656

    work page doi:10.1214/aos/1176343656 1976

  26. [26]

    , title =

    Jones, M.C. , title =. Australian Journal of Statistics , volume =. doi:https://doi.org/10.1111/j.1467-842X.1990.tb01031.x , url =. https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1467-842X.1990.tb01031.x , abstract =

    work page doi:10.1111/j.1467-842x.1990.tb01031.x 1990

  27. [27]

    2019 , eprint=

    Asymptotic Theory of L -Statistics and Integrable Empirical Processes , author=. 2019 , eprint=

    2019

  28. [28]

    2026 , eprint=

    Convergence in distribution of the P-P process in L^1[0,1] , author=. 2026 , eprint=

    2026

  29. [29]

    1986 , publisher=

    Empirical Processes with Applications to Statistics , author=. 1986 , publisher=

    1986

  30. [30]

    Terrell and David W

    George R. Terrell and David W. Scott , journal =. Variable Kernel Density Estimation , urldate =

  31. [31]

    Van der Vaart, A. W. , year=. Asymptotic Statistics , DOI=

  32. [32]

    Van der Vaart, A. W. and Wellner, J. A. , year=. Weak Convergence and Empirical Processes , publisher=

  33. [33]

    van der Vaart and Jon A

    Aad W. van der Vaart and Jon A. Wellner , title =. Asymptotics: Particles, Processes and Inverse Problems , editor =. 2007 , doi =

    2007

  34. [34]

    J. G. Wendel , journal =. Note on the Gamma Function , urldate =

  35. [35]

    Whittaker, E. T. and Watson, G. N. , year=. A Course of Modern Analysis , DOI=

  36. [36]

    Goodness-of-fit testing for

    Chhor, Julien and Carpentier, Alexandra , journal=. Goodness-of-fit testing for

  37. [37]

    Local goodness-of-fit testing for

    Chhor, Julien and Carpentier, Alexandra , journal=. Local goodness-of-fit testing for. 2025 , publisher=

    2025