pith. sign in

arxiv: 2606.19086 · v1 · pith:KPD6FNZPnew · submitted 2026-06-17 · 📊 stat.ME

Probability Bound Analysis for Dependence Uncertainty in Risk and Decision Models

Rowan Iskandar This is my paper

Pith reviewed 2026-06-26 19:51 UTC · model grok-4.3

classification 📊 stat.ME
keywords probability bound analysisdependence uncertaintyp-boxescopulasrisk modelsepistemic uncertaintyblack-box modelsFréchet bounds
0
0 comments X

The pith

Dependence uncertainty in risk models widens output bounds when propagated through probability boxes and admissible couplings.

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

The paper develops a dependence-sensitive probability bound analysis framework for black-box risk and decision models where both marginal distributions and dependence structures may be only partly known. It combines p-boxes for imprecise marginals, copulas for any fully specified dependence, and Fréchet-style admissible coupling sets for unknown dependence, including cross-dependence between imprecise and precise inputs. A sympathetic reader would care because standard PBA applications often default to independence or fixed dependence, which the illustrative example shows can produce materially narrower bounds and tail-risk summaries than the evidence actually supports. The framework therefore supplies a way to propagate all dependence structures consistent with the given marginal information without requiring a single fully justified copula.

Core claim

By propagating p-box parameters, precise-CDF parameters, and fixed quantities through arbitrary black-box models while incorporating specified dependence through copulas and unknown dependence through Fréchet admissible coupling sets, the framework yields output probability bounds that correctly reflect epistemic uncertainty arising from incomplete dependence information; the illustrative risk decision model demonstrates that varying the dependence assumptions changes both the width of the output bounds and the tail-risk summaries.

What carries the argument

The dependence-sensitive PBA framework that combines p-boxes for marginal uncertainty with copulas for known dependence and Fréchet admissible coupling sets for unknown dependence, allowing propagation through black-box models.

If this is right

  • Dependence assumptions materially affect output bounds and tail-risk summaries in risk decision models.
  • Analyses that ignore or simplify dependence produce narrower characterizations of plausible outcomes than the available evidence warrants.
  • The framework supports transparent uncertainty propagation when evidence is insufficient to justify either precise marginal distributions or a single dependence model.
  • Cross-dependence between imprecisely specified and precisely specified inputs can be incorporated without requiring full specification of all joints.

Where Pith is reading between the lines

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

  • The same coupling construction could be tested on higher-dimensional input vectors to determine whether the computational cost of enumerating admissible couplings remains tractable.
  • Existing risk assessments in finance or reliability engineering that currently assume independence might need re-computation under this framework to check whether reported tail probabilities are understated.
  • If the admissible sets prove conservative in practice, hybrid approaches that tighten them with additional qualitative dependence information could be explored.

Load-bearing premise

That Fréchet-style admissible coupling sets combined with copulas can be propagated through arbitrary black-box models while correctly representing all plausible dependence structures consistent with the given marginal information.

What would settle it

A concrete counter-example in which some dependence structure consistent with the input marginals produces an output distribution lying outside the bounds computed by the admissible-coupling construction.

read the original abstract

Risk and decision models often combine sparse marginal information, precisely specified probability distributions, and dependence assumptions that are only partly justified. Probability bound analysis (PBA) represents epistemic uncertainty through probability boxes, but many applications assume independence or require dependence structures to be fully specified. We develop a dependence-sensitive PBA framework for black-box risk and decision models in which both marginal information and dependence information may be incomplete. The framework combines p-box parameters, precise-CDF parameters, and fixed quantities; incorporates specified dependence through copulas; and propagates unknown dependence through Fr\'echet-style admissible coupling sets. We also extend the construction to cross-dependence between imprecisely specified and precisely specified inputs. In an illustrative risk decision model, dependence assumptions materially affected output bounds and tail-risk summaries; analyses that ignored or simplified dependence produced narrower characterizations of plausible outcomes. The framework supports transparent uncertainty propagation when evidence is insufficient to justify either precise marginal distributions or a single dependence model.

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

1 major / 1 minor

Summary. The paper develops a dependence-sensitive probability bound analysis (PBA) framework for black-box risk and decision models. It combines p-box parameters for epistemic uncertainty, precise-CDF parameters, and fixed quantities; incorporates specified dependence via copulas; and propagates unknown dependence through Fréchet-style admissible coupling sets. The framework is extended to cross-dependence between imprecisely and precisely specified inputs. In an illustrative risk decision model, dependence assumptions materially affect output bounds and tail-risk summaries, while analyses ignoring or simplifying dependence produce narrower characterizations of plausible outcomes.

Significance. If the propagation of Fréchet admissible couplings through general black-box models can be shown to produce tight output p-boxes representing all joints consistent with the given marginal information, the framework would offer a useful advance for uncertainty quantification in risk models where both marginals and dependence are incompletely specified. The illustrative result provides concrete evidence that dependence uncertainty can materially widen plausible outcome ranges, supporting the value of the approach for transparent analysis.

major comments (1)
  1. [Framework description and illustrative example] The central claim requires that Fréchet-style admissible coupling sets (combined with copulas) can be pushed forward through arbitrary black-box models to produce correct output p-boxes. For non-monotonic or high-dimensional black-box functions this push-forward generally requires solving a non-convex optimization over the admissible copula class; no general exact method is guaranteed to be tight or feasible without additional assumptions such as monotonicity in each argument. The illustrative example may hold only because the chosen model satisfies such hidden conditions.
minor comments (1)
  1. The abstract states the framework exists and reports an illustrative effect but provides no derivations, algorithms, or validation details, making the central claim difficult to verify from the given information.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the careful reading and for highlighting an important computational aspect of the framework. We respond to the major comment below.

read point-by-point responses

  1. Referee: [Framework description and illustrative example] The central claim requires that Fréchet-style admissible coupling sets (combined with copulas) can be pushed forward through arbitrary black-box models to produce correct output p-boxes. For non-monotonic or high-dimensional black-box functions this push-forward generally requires solving a non-convex optimization over the admissible copula class; no general exact method is guaranteed to be tight or feasible without additional assumptions such as monotonicity in each argument. The illustrative example may hold only because the chosen model satisfies such hidden conditions.

    Authors: The framework defines the output p-boxes conceptually as the pointwise infimum and supremum of the push-forwards of all joints consistent with the input marginal information (p-boxes or precise CDFs) and the admissible coupling sets (Fréchet bounds or copula-constrained). This definition is model-independent and holds for arbitrary black-box functions by construction of the admissible set. We agree, however, that realizing these bounds computationally for non-monotonic or high-dimensional black-boxes requires solving a generally non-convex optimization problem over the admissible copula class, and the manuscript does not supply a general exact algorithm guaranteed to be tight or tractable. The illustrative risk-decision model was chosen because its structure permits explicit or numerical evaluation of the extremal couplings; we do not claim that the same ease of computation extends to arbitrary models. We will revise the manuscript to (i) state explicitly that the framework supplies the theoretical bounds while practical computation is model-dependent and may require case-specific optimization or conservative approximations, and (ii) add a brief discussion of the monotonicity or low-dimensional conditions that facilitate exact computation. This clarification does not change the core contribution but addresses the referee's concern directly. revision: yes

Circularity Check

0 steps flagged

No circularity: framework combines standard p-box and copula tools without self-referential reduction.

full rationale

The abstract and description present a methodological framework that combines existing p-box parameters, precise CDFs, copulas for specified dependence, and Fréchet admissible sets for unknown dependence, then applies them to an illustrative model. No equations, fitted parameters, or self-citations are shown that would make any output equivalent to its inputs by construction. The claim that dependence assumptions affect output bounds is an empirical observation from the example rather than a definitional tautology. The derivation chain therefore remains self-contained against external benchmarks and receives the default non-circular finding.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The framework rests on standard properties of copulas and Fréchet bounds; no free parameters, invented entities, or ad-hoc axioms are stated in the abstract.

axioms (1)
  • standard math Copula representations and Fréchet admissible sets correctly capture all dependence structures consistent with given marginal information.
    Invoked to propagate both specified and unknown dependence.

pith-pipeline@v0.9.1-grok · 5682 in / 1198 out tokens · 28348 ms · 2026-06-26T19:51:35.748999+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

26 extracted references · 6 canonical work pages

  1. [1]

    Chichester: John Wiley & Sons, 2006

    O’Hagan A, Buck CE, Daneshkhah A, et al.Uncertain Judgements: Eliciting Experts’ Probabilities. Chichester: John Wiley & Sons, 2006

    2006

  2. [2]

    Troffaes MCM, Gosling JP. Robust detection of exotic infectious diseases in animal herds: A comparative study of three decision methodologies under severe uncertainty.International Journal of Approximate Reasoning.2012;53(8):1271–1281. doi: 10.1016/j.ijar.2012.06.020 30

    work page doi:10.1016/j.ijar.2012.06.020 2012

  3. [3]

    John Wiley & Sons, 2014

    Augustin T, Coolen FPA, Cooman dG, Troffaes MCM., eds.Introduction to Imprecise Probabilities. John Wiley & Sons, 2014

    2014

  4. [4]

    Chapman and Hall, 1991

    Walley P.Statistical Reasoning with Imprecise Probabilities. Chapman and Hall, 1991

    1991

  5. [5]

    Uncertainty analysis based on probability bounds analysis.Risk Analysis.2009;29(8):1156– 1169

    Ferson S, Kreinovich V , Ginzburg L, Myers DS, Sentz K. Uncertainty analysis based on probability bounds analysis.Risk Analysis.2009;29(8):1156– 1169

    2009

  6. [6]

    Dependence in Probabilistic Modeling, Dempster-Shafer Theory, and Probability Bounds Analysis

    Ferson S, Nelsen RB, Hajagos J, et al. Dependence in Probabilistic Modeling, Dempster-Shafer Theory, and Probability Bounds Analysis. Tech. Rep. SAND2004-3072, Sandia National Laboratories; Albuquerque, NM: 2004

    2004

  7. [7]

    Springer

    Nelsen RB.An Introduction to Copulas. Springer. 2 ed., 2006

    2006

  8. [8]

    Chapman and Hall/CRC, 2014

    Joe H.Dependence Modeling with Copulas. Chapman and Hall/CRC, 2014

    2014

  9. [9]

    Probabilistic sensitivity analysis of complex models: a Bayesian approach.Journal of the Royal Statistical Society: Series B.2004;66(3):751–769

    Oakley JE, O’Hagan A. Probabilistic sensitivity analysis of complex models: a Bayesian approach.Journal of the Royal Statistical Society: Series B.2004;66(3):751–769. doi: 10.1111/j.1467-9868.2004.05304.x

    work page doi:10.1111/j.1467-9868.2004.05304.x 2004

  10. [10]

    Probabilistic sensitivity analysis for NICE technology assessment: not an optional extra.Health Economics.2005;14(4):339–347

    Claxton K, Sculpher M, McCabe C, et al. Probabilistic sensitivity analysis for NICE technology assessment: not an optional extra.Health Economics.2005;14(4):339–347

    2005

  11. [11]

    Accounting for uncertainty in health economic decision models by using value of information analysis

    Jackson CH, Thompson SG, Sharples LD. Accounting for uncertainty in health economic decision models by using value of information analysis. Medical Decision Making.2019;39(5):497–509

    2019

  12. [12]

    Sensitivity analysis in presence of model uncertainty and correlated inputs , journal =

    Jacques J, Lavergne C, Devictor N. Sensitivity analysis in presence of model uncertainty and correlated inputs.Reliability Engineering & System Safety.2006;91(10-11):1126–1134. doi: 10.1016/j.ress.2005.11.047

    work page doi:10.1016/j.ress.2005.11.047 2006

  13. [13]

    Variance-based sensitivity indices for models with dependent inputs.Reliability Engineering & System Safety.2012;107:115–

    Mara TA, Tarantola S. Variance-based sensitivity indices for models with dependent inputs.Reliability Engineering & System Safety.2012;107:115–

    2012

  14. [14]

    doi: 10.1016/j.ress.2011.08.008

    work page doi:10.1016/j.ress.2011.08.008 2011

  15. [15]

    An algorithm to generate correlated input-parameters to perform probabilistic sensitivity analysis in cost-effectiveness analysis.PharmacoEconomics Open.2020;4:1–13

    Neine M, Brianc¸on S, others . An algorithm to generate correlated input-parameters to perform probabilistic sensitivity analysis in cost-effectiveness analysis.PharmacoEconomics Open.2020;4:1–13. doi: 10.1007/s41669-020-00225-8

    work page doi:10.1007/s41669-020-00225-8 2020

  16. [16]

    Fréchet-Bounds and Their Applications

    Rüschendorf L. Fréchet-Bounds and Their Applications. In: , , Springer, 1991:151–187

    1991

  17. [17]

    John Wiley & Sons, 2014

    Troffaes MCM, Cooman dG.Lower Previsions. John Wiley & Sons, 2014

    2014

  18. [18]

    Probability bound analysis: A novel approach for quantifying parameter uncertainty in decision-analytic modeling and cost-effectiveness analysis.Statistics in Medicine.2021;40:1–22

    Iskandar R. Probability bound analysis: A novel approach for quantifying parameter uncertainty in decision-analytic modeling and cost-effectiveness analysis.Statistics in Medicine.2021;40:1–22

    2021

  19. [19]

    Iskandar R. Probability bound analysis: A novel approach for quantifying parameter uncertainty in decision-analytic modeling and cost-effectiveness analysis.Statistics in Medicine.2021;40(29):6501–6522

    2021

  20. [20]

    Probabilistic sensitivity analysis in health economics.Statistical Methods in Medical Research.2011;24(6):615–634

    Baio G, Dawid AP. Probabilistic sensitivity analysis in health economics.Statistical Methods in Medical Research.2011;24(6):615–634. doi: 10.1177/0962280211419832

    work page doi:10.1177/0962280211419832 2011

  21. [21]

    Active learning for efficiently training emulators of computationally expensive mathematical models.Statistics in Medicine.2020;39(25):3521–3548

    Ellis AG, Iskandar R, Schmid CH, Wong JB, Trikalinos TA. Active learning for efficiently training emulators of computationally expensive mathematical models.Statistics in Medicine.2020;39(25):3521–3548

    2020

  22. [22]

    Princeton University Press, 1944

    Neumann vJ, Morgenstern O.Theory of Games and Economic Behavior. Princeton University Press, 1944

    1944

  23. [23]

    John Wiley & Sons, 1954

    Savage LJ.The F oundations of Statistics. John Wiley & Sons, 1954

    1954

  24. [24]

    The irrelevance of inference: a decision-making approach to the stochastic evaluation of health care technologies.Journal of Health Economics.1999;18(3):341–364

    Claxton K. The irrelevance of inference: a decision-making approach to the stochastic evaluation of health care technologies.Journal of Health Economics.1999;18(3):341–364

    1999

  25. [25]

    Representing uncertainty: the role of cost-effectiveness acceptability curves.Health Economics.2001;10(8):779– 787

    Fenwick E, Claxton K, Sculpher M. Representing uncertainty: the role of cost-effectiveness acceptability curves.Health Economics.2001;10(8):779– 787

    2001

  26. [26]

    Maxmin expected utility with non-unique prior.Journal of Mathematical Economics.1989;18(2):141–153

    Gilboa I, Schmeidler D. Maxmin expected utility with non-unique prior.Journal of Mathematical Economics.1989;18(2):141–153

    1989