Newsvendor under Ambiguity and Misspecification

Feng Liu; Ruodu Wang; Shuming Wang; Zhi Chen

arxiv: 2405.07008 · v3 · submitted 2024-05-11 · 🧮 math.OC

Newsvendor under Ambiguity and Misspecification

Feng Liu , Zhi Chen , Ruodu Wang , Shuming Wang This is my paper

Pith reviewed 2026-05-24 01:17 UTC · model grok-4.3

classification 🧮 math.OC

keywords newsvendor problemambiguity aversionmisspecificationoptimal transportmean-variance ambiguity setclosed-form solutionfinite-sample guaranteeinventory optimization

0 comments

The pith

A closed-form optimal order quantity for the newsvendor generalizes the Scarf model by incorporating both ambiguity and misspecification aversion.

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

The paper addresses the newsvendor problem when the demand distribution is unknown, separating ambiguity over distributions with shared characteristics from misspecification of those characteristics. It introduces a criterion that maximizes worst-case expected profit regularized by the optimal transport distance to a mean-variance ambiguity set. This yields a closed-form optimal order quantity extending the Scarf solution under ambiguity alone. Finite-sample performance guarantees are established that account for estimation error and distribution shift. The results show that misspecification aversion can reverse the effect of higher prices or variance on the optimal quantity.

Core claim

We derive the closed-form optimal order quantity that generalizes the solution of the Scarf model under only ambiguity aversion. The decision criterion of misspecification aversion possesses insightful interpretations as distributional transforms. We establish the finite-sample performance guarantee, which consists of two parts: in-sample optimal value and out-of-sample effect of misspecification that can be further decoupled into estimation error and distribution shift. The closed-form solution highlights the impact of misspecification aversion: the optimal order quantity under misspecification aversion can decrease as the price or variance increases, reversing the monotonicity of that only

What carries the argument

The misspecification-averse decision criterion that regularizes worst-case expected profit by the optimal-transport distance of a distribution to the mean-variance ambiguity set.

If this is right

The optimal order quantity is available in closed form and generalizes the Scarf model.
It can decrease as price or variance increases, reversing the behavior under ambiguity aversion alone.
Finite-sample performance guarantees hold, separating in-sample value from out-of-sample misspecification effects.
The framework extends to multiple products and alternative distributional distances like total variation.

Where Pith is reading between the lines

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

The reversal in monotonicity suggests that operational decisions under layered uncertainty require separate treatment of misspecification.
This regularization approach may yield analogous closed-form solutions in other single-period stochastic inventory models.
Real-world non-stationary demand data could be used to test whether the decoupled performance guarantee holds in practice.

Load-bearing premise

The regularization using optimal-transport cost to the mean-variance ambiguity set accurately models the joint effect of ambiguity and misspecification aversion.

What would settle it

Computing the closed-form order quantity for concrete mean-variance parameters and prices and finding that it increases rather than decreases with price would falsify the derived solution.

read the original abstract

Problem definition: We consider a newsvendor problem with unknown demand distribution, where we distinguish ambiguity under which the newsvendor does not differentiate demand distributions of common characteristics and misspecification under which such characteristics might be misspecified. Methodology/results: The newsvendor hedges against ambiguity and misspecification by maximizing the worst-case expected profit regularized by a distribution's distance to an ambiguity set. Focusing on the popular mean-variance ambiguity set and optimal-transport cost for the misspecification, we show that the decision criterion of misspecification aversion possesses insightful interpretations as distributional transforms. We derive the closed-form optimal order quantity that generalizes the solution of the Scarf model under only ambiguity aversion. We establish the finite-sample performance guarantee, which consists of two parts: in-sample optimal value and out-of-sample effect of misspecification that can be further decoupled into estimation error and distribution shift. We also extend the framework to multiple products, distributional characteristics specified via optimal transport, and misspecification measured by total variation distance. Managerial implications: The closed-form solution highlights the impact of misspecification aversion: the optimal order quantity under misspecification aversion can decrease as the price or variance increases, reversing the monotonicity of that under only ambiguity aversion. Hence, ambiguity and misspecification, as different layers of distributional uncertainty, can result in distinct operational consequences. The finite-sample performance guarantee theoretically justifies the necessity of incorporating misspecification aversion in a non-stationary environment, which is also well demonstrated in our experiments with real-world data.

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

The paper adds an optimal-transport misspecification penalty to the mean-variance ambiguous newsvendor and obtains a closed-form order quantity plus decoupled finite-sample bounds.

read the letter

The core contribution is a regularization of worst-case profit that penalizes distance to the ambiguity set, producing an explicit solution that extends Scarf and can reverse the usual monotonicity in price and variance. They also give finite-sample guarantees that split estimation error from distribution shift, plus extensions to multi-product cases and total-variation misspecification. That separation of concerns and the explicit reversal are the parts that stand out as new relative to the cited literature. The distributional-transform interpretation is a clean way to read the criterion. The finite-sample result looks like the kind of guarantee that could be useful in non-stationary settings. The abstract states the claims directly without obvious circularity or hidden non-convexity. The main limitation is that the derivations, any post-hoc parameter choices, and the real-data experiments are not visible here, so the tightness of the bounds and the practical size of the reversal remain unverified. The mean-variance focus is standard but narrow; whether the same closed form survives other ambiguity sets is left open. This is squarely for operations-research readers working on robust inventory models. Someone already citing Scarf or distributionally robust newsvendor work would see the incremental step and the managerial distinction between ambiguity and misspecification. It is coherent on its own terms and specific enough to merit referee time, even if the proofs need checking.

Referee Report

2 major / 3 minor

Summary. The manuscript addresses the newsvendor problem under two layers of distributional uncertainty: ambiguity (indistinguishability within a mean-variance set) and misspecification (possible errors in characteristics). The decision criterion regularizes worst-case expected profit by an optimal-transport penalty on distance to the ambiguity set. For the mean-variance case the authors derive a closed-form optimal order quantity that generalizes the Scarf solution, establish finite-sample guarantees that decouple in-sample optimality from out-of-sample misspecification effects (further split into estimation error and distribution shift), and extend the framework to multi-product settings, OT-specified characteristics, and total-variation misspecification. Managerial implications highlight that misspecification aversion can reverse the monotonicity of the optimal order quantity with respect to price or variance.

Significance. If the derivations hold, the work supplies a tractable, explicit solution for inventory decisions under layered uncertainty together with finite-sample performance bounds that separate estimation and shift effects. The closed-form result, the decoupled guarantee, and the demonstration of reversed monotonicity are concrete strengths. Real-data experiments are noted as supporting the non-stationary justification. These elements would be useful additions to the distributionally robust optimization literature in operations research.

major comments (2)

[§4, Theorem 1] §4, Theorem 1 (closed-form order quantity): the derivation must explicitly track how the optimal-transport regularization term modifies the Scarf critical fractile; if the resulting expression remains closed-form only because the mean-variance set and OT cost admit a specific quadratic structure, this dependence should be stated as a modeling assumption rather than presented as a general extension.
[§5.2] §5.2, finite-sample guarantee theorem: the claimed decoupling of the out-of-sample effect into independent estimation-error and distribution-shift terms requires an explicit statement of the conditions (e.g., bounded moments, fixed ambiguity-set radius) under which the two terms do not interact; without this the guarantee cannot be applied directly to non-stationary data as asserted in the managerial implications.

minor comments (3)

The interpretation of the regularized criterion as 'distributional transforms' is mentioned in the abstract but receives no concrete illustration until later sections; a short example immediately after the problem formulation would improve readability.
Notation for the misspecification regularization parameter is introduced without a dedicated symbol table; consistent use of a single symbol (e.g., λ) across theorems and experiments would reduce confusion.
[multi-product extension] In the multi-product extension, the statement that the problem 'remains tractable' should be accompanied by the precise complexity (e.g., still O(1) per product or requires solving a small LP).

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments and positive assessment of the manuscript. Below we respond point-by-point to the major comments, indicating the revisions we will incorporate.

read point-by-point responses

Referee: [§4, Theorem 1] §4, Theorem 1 (closed-form order quantity): the derivation must explicitly track how the optimal-transport regularization term modifies the Scarf critical fractile; if the resulting expression remains closed-form only because the mean-variance set and OT cost admit a specific quadratic structure, this dependence should be stated as a modeling assumption rather than presented as a general extension.

Authors: We agree that the closed-form expression in Theorem 1 arises from the quadratic structure of both the mean-variance ambiguity set and the optimal-transport cost. In the revised version we will expand the proof of Theorem 1 to explicitly isolate each step at which the OT penalty modifies the Scarf critical fractile, and we will add a remark clarifying that the closed-form result is tied to this quadratic structure while the broader modeling framework (ambiguity plus misspecification aversion) is not. revision: yes
Referee: [§5.2] §5.2, finite-sample guarantee theorem: the claimed decoupling of the out-of-sample effect into independent estimation-error and distribution-shift terms requires an explicit statement of the conditions (e.g., bounded moments, fixed ambiguity-set radius) under which the two terms do not interact; without this the guarantee cannot be applied directly to non-stationary data as asserted in the managerial implications.

Authors: We will revise the statement of the finite-sample guarantee (Theorem 2) to list the required conditions—bounded second moments and a fixed ambiguity-set radius—under which the estimation-error and distribution-shift terms decouple. We will also add a short paragraph in the managerial-implications section explaining how these conditions support application to non-stationary environments. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper derives a closed-form optimal order quantity via regularization of worst-case expected profit using optimal-transport cost over a mean-variance ambiguity set, generalizing the Scarf model. This is presented as a direct optimization result with finite-sample guarantees decoupled into in-sample value and out-of-sample effects. No steps reduce by construction to fitted inputs, self-definitions, or load-bearing self-citations; the central claims rest on explicit optimization and theoretical analysis rather than renaming or smuggling prior results. The derivation is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The framework rests on modeling choices for the ambiguity set and distance measure plus standard optimization assumptions; no free parameters are explicitly named in the abstract but the regularization strength is implicitly present.

free parameters (1)

misspecification regularization parameter
Controls the weight on distance to the ambiguity set; its value is required to obtain the closed-form solution but is not specified as data-driven or fixed in the abstract.

axioms (2)

domain assumption Mean-variance ambiguity set adequately captures common characteristics of demand distributions
The paper focuses on this popular set without deriving its suitability from first principles.
domain assumption Optimal transport cost is an appropriate metric for measuring misspecification
Used to penalize distance to the ambiguity set; choice is stated but not justified beyond popularity.

pith-pipeline@v0.9.0 · 5809 in / 1383 out tokens · 23544 ms · 2026-05-24T01:17:40.998500+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We derive the closed-form optimal order quantity that generalizes the solution of the Scarf model under only ambiguity aversion... min_F {E_F[π(q,u)] + α d(F,A)} with d via optimal-transport quadratic cost
IndisputableMonolith/Foundation/AlphaCoordinateFixation.lean costAlphaLog_high_calibrated_iff unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Transform function φ_α(v) = α/p v² (or piecewise) yielding T_φ[G]

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

16 extracted references · 16 canonical work pages

[1]

Besbes, Omar, Will Ma, Omar Mouchtaki. 2023. From contextual data to newsvendor decisions: On the actual performance of data-driven algorithms. Available at https://arxiv.org/abs/2302.08424

work page arXiv 2023
[2]

Dowson, D.C., B.V. Landau. 1982. The Fr \'e chet distance between multivariate normal distributions. Journal of Multivariate Analysis 12(3) 450--455

work page 1982
[3]

Gao, Rui, Anton Kleywegt. 2023. Distributionally robust stochastic optimization with Wasserstein distance. Mathematics of Operations Research 48(2) 603–655

work page 2023
[4]

Gelbrich, Matthias. 1990. On a formula for the L^2 Wasserstein metric between measures on Euclidean and Hilbert spaces. Mathematische Nachrichten 147(1) 185--203

work page 1990
[5]

Wallace, Steve Zymler

Hanasusanto, Grani A., Daniel Kuhn, Stein W. Wallace, Steve Zymler. 2015. Distributionally robust multi-item newsvendor problems with multimodal demand distributions. Mathematical Programming 152(1) 1--32

work page 2015
[6]

Kupiec, Paul. 2002. Stress testing in a value at risk framework. Risk management: value at risk and beyond , 76--99. Cambridge University Press

work page 2002
[7]

M \"u ller, Alfred. 1997. Integral probability metrics and their generating classes of functions. Advances in Applied Probability 29(2) 429--443

work page 1997
[8]

Natarajan, Karthik, Chung-Piaw Teo. 2017. On reduced semidefinite programs for second order moment bounds with applications. Mathematical Programming 161 487--518

work page 2017
[9]

Natarajan, Karthik, Melvyn Sim, Joline Uichanco. 2018. Asymmetry and ambiguity in newsvendor models. Management Science 64(7) 3146--3167

work page 2018
[10]

Rigollet, Philippe, H \"u tter Jan-Christian. 2023. High-dimensional statistics. Available at https://arxiv.org/abs/2310.19244

work page arXiv 2023
[11]

Scarf, Herbert. 1958. A min-max solution of an inventory problem. Studies in the mathematical theory of inventory and production

work page 1958
[12]

Villani, C \'e dric. 2009. Optimal transport: Old and new . Springer

work page 2009
[13]

Wainwright, Martin. 2019. High-dimensional statistics: A non-asymptotic viewpoint. Cambridge University Press

work page 2019
[14]

Zhang, Luhao, Jincheng Yang, Rui Gao. 2025. A short and general duality proof for Wasserstein distributionally robust optimization. Operations Research 73(4) 2146--2155

work page 2025
[15]

, " * write output.state after.block = add.period write newline

ENTRY address author booktitle chapter doi edition editor eid howpublished institution journal key month note number organization pages publisher school series title type url volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.all := #1 'mid.sent...

work page
[16]

write newline

" write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in "" FUNCTION format.date year ...

work page

[1] [1]

Besbes, Omar, Will Ma, Omar Mouchtaki. 2023. From contextual data to newsvendor decisions: On the actual performance of data-driven algorithms. Available at https://arxiv.org/abs/2302.08424

work page arXiv 2023

[2] [2]

Dowson, D.C., B.V. Landau. 1982. The Fr \'e chet distance between multivariate normal distributions. Journal of Multivariate Analysis 12(3) 450--455

work page 1982

[3] [3]

Gao, Rui, Anton Kleywegt. 2023. Distributionally robust stochastic optimization with Wasserstein distance. Mathematics of Operations Research 48(2) 603–655

work page 2023

[4] [4]

Gelbrich, Matthias. 1990. On a formula for the L^2 Wasserstein metric between measures on Euclidean and Hilbert spaces. Mathematische Nachrichten 147(1) 185--203

work page 1990

[5] [5]

Wallace, Steve Zymler

Hanasusanto, Grani A., Daniel Kuhn, Stein W. Wallace, Steve Zymler. 2015. Distributionally robust multi-item newsvendor problems with multimodal demand distributions. Mathematical Programming 152(1) 1--32

work page 2015

[6] [6]

Kupiec, Paul. 2002. Stress testing in a value at risk framework. Risk management: value at risk and beyond , 76--99. Cambridge University Press

work page 2002

[7] [7]

M \"u ller, Alfred. 1997. Integral probability metrics and their generating classes of functions. Advances in Applied Probability 29(2) 429--443

work page 1997

[8] [8]

Natarajan, Karthik, Chung-Piaw Teo. 2017. On reduced semidefinite programs for second order moment bounds with applications. Mathematical Programming 161 487--518

work page 2017

[9] [9]

Natarajan, Karthik, Melvyn Sim, Joline Uichanco. 2018. Asymmetry and ambiguity in newsvendor models. Management Science 64(7) 3146--3167

work page 2018

[10] [10]

Rigollet, Philippe, H \"u tter Jan-Christian. 2023. High-dimensional statistics. Available at https://arxiv.org/abs/2310.19244

work page arXiv 2023

[11] [11]

Scarf, Herbert. 1958. A min-max solution of an inventory problem. Studies in the mathematical theory of inventory and production

work page 1958

[12] [12]

Villani, C \'e dric. 2009. Optimal transport: Old and new . Springer

work page 2009

[13] [13]

Wainwright, Martin. 2019. High-dimensional statistics: A non-asymptotic viewpoint. Cambridge University Press

work page 2019

[14] [14]

Zhang, Luhao, Jincheng Yang, Rui Gao. 2025. A short and general duality proof for Wasserstein distributionally robust optimization. Operations Research 73(4) 2146--2155

work page 2025

[15] [15]

, " * write output.state after.block = add.period write newline

ENTRY address author booktitle chapter doi edition editor eid howpublished institution journal key month note number organization pages publisher school series title type url volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.all := #1 'mid.sent...

work page

[16] [16]

write newline

" write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in "" FUNCTION format.date year ...

work page