Pith. sign in

REVIEW 3 major objections 4 minor 41 references

Marginal Fairness: Fair Decision-Making under Risk Measures

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper proves that any risk-measure decision has a unique minimally adjusted fair version, given by an explicit correction to the distortion weight.

desk verdict The fairness-at-the-decision-layer framing is new and worth reading, but Theorem 4.2's closed-form weight lives outside the admissible function class; the main result needs repair. read the letter →

arxiv 2505.18895 v1 pith:MZWOQI3M submitted 2025-05-24 stat.ML cs.CCcs.CYcs.LGq-fin.RM

Fei Huang , Silvana M. Pesenti This is my paper · ORCID
classification stat.MLcs.CCcs.CYcs.LGq-fin.RM MSC 91B3062P05
keywords marginalfairnessgeneralizeddistortionriskmeasuresindividualinsurancepricingsensitivityanalysiscascadeprotectedattributesExpectedShortfall
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

The paper tries to establish that fairness in risk-sensitive decisions, such as insurance premiums and capital loadings, can be enforced at the decision stage rather than the prediction stage by making the decision rule insensitive to small perturbations of protected attributes. It defines this as marginal fairness, derives a closed-form sensitivity for generalized distortion risk measures, and proves that a unique minimally adjusted fair decision rule exists and is given by an explicit correction to the distortion weight. This matters because regulated industries may legally use protected attributes in modeling but not in final decisions, and simply removing those attributes leaves indirect discrimination through correlated covariates. The paper claims the framework covers continuous, bounded, discrete, categorical, and multivariate protected attributes, and demonstrates on auto insurance data that the fair rule preserves segmentation power.

What carries the argument

Three objects carry the argument. First, the differential sensitivity $\partial_{D_i}\rho_\gamma(Y|X)$, the Gâteaux derivative of the decision rule under an infinitesimal perturbation of $D_i$; the perturbation is multiplicative $D_i(1+\delta)$ for real-valued covariates, a normal-latent distortion $F_{D_i}^{-1}(\Phi(\Phi^{-1}(U)(1+\delta)))$ for bounded or discrete covariates, and an inverse-Rosenblatt cascade for dependent covariates. Second, the generalized distortion weight $\gamma$ acting on the conditional rank $U_{Y|X}=F_{Y|X}(Y|X)$, which converts the sensitivity into an expectation of $D_i\partial_i g(D,X)\gamma(U_{Y|X})$ conditional on $X$. Third, the $L^2$ projection of $\gamma$ onto the linear constraint $\partial_{D_i}\rho_\ell(Y|X)=0$, which yields the closed-form Lagrange multiplier in Theorem 4.2 and the explicit fair rule.

What would settle it

On a simple linear model $Y = \beta_0 + \beta_1 X + \beta_2 D + \varepsilon$ with continuous $D$, compute the marginally fair expected value under the paper's multiplicative perturbation $D(1+\delta)$; the sensitivity is $\beta_2\mathbb{E}[D|X]$. Repeat the derivation with an additive perturbation $D+\delta$: the sensitivity becomes $\beta_2$, so the 'fair' premium produced by the same optimization changes. Estimating both versions of the fair premium on a real dataset would directly show that the fairness guarantee depends on the chosen perturbation direction.

Watch

Extended reading notes

Core claim

The paper claims that for a decision rule $\rho_\gamma(Y|X)$ built from a generalized distortion risk measure with square-integrable weight $\gamma$, enforcing marginal fairness with respect to a protected covariate $D_i$ has a unique solution. The fair distortion weight is $$\gamma^*(U_{Y|X}) = \gamma(U_{Y|X}) - \frac{\partial_{D_i} \rho_\gamma(Y|X)}{\mathbb{E}[(D_i \partial_i g(D,X))^2 | X]} \, D_i \, \partial_i g(D,X), \quad P\text{-a.s.},$$ and the corresponding fair decision is $$\$rho^{{D_i}}$_{\gamma^*}(Y|X) = \rho_\gamma(Y|X) - \frac{\partial_{D_i}\rho_\gamma(Y|X)}{\mathbb{E}[(D_i\partial_i g(D,X))^2|X]}\, \mathbb{E}[Y D_i \partial_i g(D,X)|X].$$ The correction removes the sensitivity of the risk measure to infinitesimal perturbations of $D_i$ while staying closest to the original weight in squared $L^2$ distance, and the same structural result is claimed to hold for bounded, discrete, categorical, and multivariate protected variables, as well as for cascade perturbations that propagate through dependent covariates.

Load-bearing premise

The whole construction depends on the chosen infinitesimal perturbation direction for each protected attribute; a different equally small perturbation can change both the measured sensitivity and the fair rule, and the paper gives no external argument that the chosen direction is the legally or ethically relevant one.

Editorial extensions

If this is right

  • Every generalized distortion risk measure, including Expected Shortfall and deviation measures, can be made marginally fair by subtracting one explicit correction term from the original decision rule.
  • The correction is computed without retraining the prediction function $g$; it only needs estimates of the sensitivity and two conditional moments.
  • Fairness extends from expected-value premiums to risk margins and capital loadings, covering the part of insurance pricing that mean-based fairness rules ignore.
  • Cascade sensitivity makes the framework responsive to indirect discrimination through correlated covariates, without requiring causal assumptions or counterfactual models.
  • In the empirical auto-insurance study, the marginally fair rule keeps Gini-based segmentation power and calibration close to the unaware and discrimination-free benchmarks.

Reading between the lines

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

  • The same $L^2$-projection argument would extend to any law-invariant functional with a square-integrable kernel and a linear sensitivity in the weight, so the theorem is probably not restricted to distortion risk measures; that is an editorial extrapolation.
  • Marginal fairness does not imply group fairness: a rule can have zero derivative with respect to $D_i$ yet still exhibit large average outcome differences across groups, because the derivative is local. A demographic-parity check on the same dataset would settle whether the two notions diverge.
  • Because the discrete perturbation uses a normal latent and order-preserving thresholds, permuting the category labels of a categorical variable changes the sensitivity coefficients $v_k$; an order-invariant fair rule would need a different perturbation design.
  • A practical monitoring test follows directly: after deploying $\rho_{\gamma^*}$, re-estimate its sensitivity on a validation stream; nonzero values indicate model drift or misspecification of $g$.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper proposes "marginal fairness," a derivative-based individual fairness criterion for decisions made with generalized distortion risk measures. The authors model decision-making as a two-step process (prediction followed by risk-measure-based pricing), define sensitivity of the decision to infinitesimal perturbations of protected covariates, and solve an L2-minimal adjustment of the distortion weight to enforce zero sensitivity (Theorem 4.2). They extend the construction to bounded, discrete, categorical, and multivariate protected variables, introduce cascade sensitivity to propagate perturbations through dependent covariates, and report a numerical study and an empirical study on French auto insurance data.

Significance. If the central theorem were correct, the paper would offer an explicit, operational formula for fair risk-measure-based decisions in regulated industries, with unusual breadth across continuous, discrete, and categorical protected attributes. The empirical section is clearly described and shows a plausible workflow. However, the main mathematical claim is not established: the proposed adjustment is generally not an admissible distortion weight, so the existence/uniqueness result fails in generic settings. This undermines the theoretical foundation and the derived corollaries for all covariate types.

major comments (3)
  1. [Theorem 4.2 and Appendix C, Eq. (C.2)-(4.1)] The pointwise Lagrangian solution is not a single weight function on (0,1). Equation (C.2) gives ℓ*(U_{g(t,x)|X}) = γ(U_{g(t,x)|X}) − η(x) t_i ∂_i g(t,x) for each (t,x), but for this to define γ* ∈ Γ the right-hand side must be a function of U_{g(t,x)|X} alone. In general it also depends on X and on the realization of D_i through t_i. Since U_{Y|X} is independent of X, a deterministic function γ*(U) cannot reproduce the claimed equality (4.1). A concrete counterexample satisfying Assumption 4.1 is X ~ U[-1,1], D | X = x ~ N(x,1), Y = arctan(D) + X, and γ ≡ 1 (expectation). Then U_{Y|X} = Φ(D−X), so D = X + Φ^{-1}(U), and the right-hand side of (4.1) becomes 1 − η(X)(X + Φ^{-1}(U))/(1 + (X + Φ^{-1}(U))^2), which depends on X and is not a function of U alone. Thus the existence and uniqueness statement of Theorem 4.2 is internally inconsistent; the proofs of Corollaries 4.7, 4.11, and 5.4 inherit the same gap.
  2. [Optimization problem (P), Section 4] The optimization problem is ill-posed as written. The class Γ_{t,x} is described as square-integrable weight functions parametrized by (t,x), but the objective ∫_0^1 (γ(u)−ℓ(u))^2 du suggests ℓ is a single function of u only, independent of (t,x). The proof of Theorem 4.2 optimizes pointwise in (t,x), which yields a different function for each x and t. If ℓ is allowed to depend on (t,x), then the decision rule ρ_ℓ(Y|X) depends on the protected attribute through the realized D, contradicting the requirement that decisions use only X. If ℓ is restricted to functions of u only, the pointwise solution is not admissible, as shown in the first major comment. The manuscript needs to specify the admissible class unambiguously.
  3. [Definition 5.2 and Theorem 5.3] Cascade sensitivity is defined with unconditional risk measures: ∂^c_{D_i} ργ(Y|X) := lim_{δ↓0} [ργ(g((D,X)_δ)) − ργ(g(D,X))]/δ, with no conditioning on X. Yet Theorem 5.3 and Example 5.5 state conditional representations with |X and use conditional expectations. If the intended definition is unconditional, the conditional formula does not follow from the definition; if the intended definition is conditional, the definition and the surrounding text must condition on X consistently. This inconsistency propagates to Proposition 5.4 and to the cascade numerical and empirical examples.
minor comments (4)
  1. [Section 1.2, last sentence] The phrase "which are all delegated go Appendix C" contains a typo; it should read "delegated to Appendix C."
  2. [Theorem 4.10 statement] The summation notation "KX" appears where the upper limit should likely be "K−1" or "K"; the same symbol is used inconsistently in the displayed formula after Eq. (4.10).
  3. [Example 4.3, Eq. (4.4)] The formula for β̄0(x) contains ργ(Y | X = x) in addition to the sensitivity term; since the displayed expression is intended to define the fair decision rule, the reader should be told explicitly that ργ(Y | X = x) here is the original, unadjusted risk measure, not the final fair rule.
  4. [Figure 12 caption] The caption describes "green squares" for the marginally fair decision under the expected value risk measure, but the legend order in the text and the colors may confuse readers; please verify the color-key correspondence.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the fair decision rule is a constrained projection constructed from the stated fairness definition, and the self-cited sensitivity results are re-proved or extended in the appendices.

full rationale

The paper's central result (Theorem 4.2) solves an explicit optimization problem (P): it minimizes the squared L2 distance between the original distortion weight and a candidate weight, subject to the marginal-fairness constraint that the derivative sensitivity vanishes. The solution is a mathematical projection onto a constraint set, not a fitted prediction, so the resulting decision rule satisfying the fairness constraint is the intended construction rather than a circular derivation. No constants are fitted to make the theorem true, and the empirical section tests the constructed decision rules against out-of-sample insurance data from CASdatasets using Gini curves and quantile plots, giving an external benchmark for the practical claims. The paper does cite prior work by the same author group for sensitivity formulas, notably [27], [28], and [29]; however, Proposition 3.4 is proved in Appendix C, the discrete-case Lemmas B.1-B.3 are proved in the paper itself, and Theorem 5.3 is presented as an application of Proposition 3.4 with a citation to a published result under stronger assumptions. These self-citations are therefore not load-bearing in the sense of replacing an argument with an unverified assertion. A possible gap in Theorem 4.2, namely whether the pointwise Lagrangian solution is well-defined as a function of U_{Y|X} alone, is a correctness concern rather than a circularity, and is not scored here.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The theoretical claim does not need fitted parameters. The axioms are the two-step decision structure, the chosen perturbation directions, the integrability assumptions, and the Rosenblatt cascade model. No new physical or mechanistic entities are introduced; the 'marginally fair rule' is a constructed decision procedure, not an entity.

free parameters (1)
  • Empirical model coefficients (Tweedie GLM and NN weights) = Estimated on training data
    Used in Section 7 to compute sensitivities and the marginal fairness correction. They do not appear in the theoretical theorem, but the empirical demonstration depends on them, and no error bars are given.
assumptions (5)
  • domain assumption The decision process is two-step: prediction g(D,X) followed by risk measure conditional only on X.
    Section 2.2, Definition 2.4. This separation is the basis for distinguishing direct from indirect discrimination and for aligning with industry practice.
  • ad hoc to paper Marginal fairness is defined as zero Gâteaux derivative with respect to a hand-chosen perturbation of the protected covariate.
    Definition 3.1 and the discussion after it. Different perturbation directions yield different fair rules, and the paper offers no external justification for one direction over another.
  • domain assumption Assumption 4.1: g is invertible in the protected component and certain second moments are bounded away from zero.
    Needed for the existence and uniqueness result in Theorem 4.2 and to ensure the adjusted weight is square-integrable.
  • ad hoc to paper For discrete and categorical covariates, the generalized distributional transform and the normal-latent perturbation (4.10) are the appropriate perturbation.
    Subsection 4.3, formula (4.10). This perturbation is a modeling choice that preserves the support of the discrete variable but is not derived from first principles.
  • domain assumption Cascade sensitivity propagates a perturbation through the joint distribution via the inverse Rosenblatt transform, without causal assumptions.
    Section 5, equations (5.1)-(5.2). This is a specific model of how dependence transmits the effect of a protected attribute to correlated covariates.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Marginal Fairness: Fair Decision-Making under Risk Measures." pith.science (2026). https://pith.science/paper/MZWOQI3M

@misc{pith2026250518895,
  author       = {Pith},
  title        = {Pith review of: Marginal Fairness: Fair Decision-Making under Risk Measures},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MZWOQI3M}},
  note         = {Machine review of arXiv:2505.18895}
}
read the original abstract

This paper introduces marginal fairness, a new individual fairness notion for equitable decision-making in the presence of protected attributes such as gender, race, and religion. This criterion ensures that decisions based on generalized distortion risk measures are insensitive to distributional perturbations in protected attributes, regardless of whether these attributes are continuous, discrete, categorical, univariate, or multivariate. To operationalize this notion and reflect real-world regulatory environments (such as the EU gender-neutral pricing regulation), we model business decision-making in highly regulated industries (such as insurance and finance) as a two-step process: (i) a predictive modeling stage, in which a prediction function for the target variable (e.g., insurance losses) is estimated based on both protected and non-protected covariates; and (ii) a decision-making stage, in which a generalized distortion risk measure is applied to the target variable, conditional only on non-protected covariates, to determine the decision. In this second step, we modify the risk measure such that the decision becomes insensitive to the protected attribute, thus enforcing fairness to ensure equitable outcomes under risk-sensitive, regulatory constraints. Furthermore, by utilizing the concept of cascade sensitivity, we extend the marginal fairness framework to capture how dependencies between covariates propagate the influence of protected attributes through the modeling pipeline. A numerical study and an empirical implementation using an auto insurance dataset demonstrate how the framework can be applied in practice.

Figures

Figures reproduced from arXiv: 2505.18895 by the authors.

Figure 1
Figure 1. A graphical representation of the decision process (arrows indicate statistical or functional dependence, not causality). The predicted outcome Y is modeled as a function of both protected attributes D and non-protected covariates X. The decision ργ(Y | X) is a function of the conditional distribution of Y given X, which is modeled as a function of only X. The dashed arrow between D and X indicates that dependence b… view at source ↗
Figure 2
Figure 2. Sensitivity to D under the assumption that D ∼ Ber(p) and independent of X from Example 4.12. The sensitivity is given in (4.13) with β2 = 1. The x-axis is the success rate, i.e. p ∈ (0, 1). 5. Marginal fairness with cascade sensitivity. In the presence of statistical dependence between the protected attributes D and the remaining covariates X, a perturbation of a sin￾gle protected feature Di may influence other com… view at source ↗
Figure 3
Figure 3. Cascading perturbation of Xδ due to a perturbation of D ∼ Bern(p) from Example 5.1. Blue lines correspond to p = 0.8 and red lines to p = 0.2. Solid lines are δ = 0 and dashed lines are the perturbation δ = 0.2. We say that a decision rule ργ is marginally fair with cascade sensitivity, if the cascade sensitivity to covariate Di vanishes. Similarly, we say the decision rule is multi-marginally fair with cascade sens… view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Comparison of fair decision strategies. Unaware decision with expected value (black), discrimination-free with expected value (grey), and marginally fair with expected value (blue) and ES (red) [PITH_FULL_IMAGE:figures/full_fig_p025_4.png]
Figure 5
Figure 5. Figure 5: Coefficient adjustment 1 − cx and 1 − c¯x for marginally fair decisions under expected value and ES, respectively [PITH_FULL_IMAGE:figures/full_fig_p025_5.png]
Figure 6
Figure 6. Figure 6: Comparison of fair decision strategies based on expected value. Marginally fair decision with marginal sensitivity (blue) and marginally fair decision with cascade sensitivity (orange) [PITH_FULL_IMAGE:figures/full_fig_p027_6.png]
Figure 7
Figure 7. Figure 7: Comparison of sensitivity of marginally fair decisions with marginal sensitivity (blue) and cascade sensitivity (orange) based on expected value. For this application, we first assume that the decision rule is based on the expected value of the outcome. This choice fac…
Figure 8
Figure 8. Figure 8: Comparison of fair decision strategies for 50 randomly selected policyholders under the expected value risk measure. Blue dots correspond to marginally fair decision, orange crosses to unaware decision, and the greed squares to the discrimination-free decision. lower e…
Figure 9
Figure 9. Figure 9: Left: Box plots of the sensitivity of the decision rule for different age groups. Right: average sensitivity of the decision rule for age groups and by gender. Both are under the expected value risk measure. cohorts where the risk of indirect discrimination is more pro…
Figure 10
Figure 10. Figure 10: presents the Gini curves for the three decision rules. The marginally fair decision rule demonstrates segmentation power on par with both the unaware and discrimination-free rules, indicating that fairness can be enforced without sacrificing risk differentiation—a key…
Figure 11
Figure 11. Figure 11: Quantile plot comparing predicted (blue) and observed (orange) losses across decision rules under the expected value risk measure. The x-axis shows exposure-weighted quantiles of predicted claim costs, and the y-axis reports average observed and predicted losses withi…
Figure 12
Figure 12. Figure 12: Comparison of fair decision strategies for 50 randomly selected policyholders under ES. Blue dots correspond to marginally fair decisions under ES, orange crosses to unaware decision under ES, and the green squares to the marginally fair decision under the expected va…
Figure 13
Figure 13. Figure 13: Left: Box plots of the sensitivity of the decision rule for different age groups. Right: average sensitivity of the decision rule for age groups and by gender. Both under ES. Statistic Min 25% Quantile 50% Quantile 75% Quantile Max PU − PMF 0.06 1.79 3.28 6.37 204.82 …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 39 canonical work pages

  1. [1]

    C. A. Araiza Iturria, M. Hardy, and P. Marriott , A discrimination-free premium under a causal framework, North American Actuarial Journal, 28 (2024), pp. 801–821. 43

    work page 2024

  2. [2]

    Barocas, M

    S. Barocas, M. Hardt, and A. Narayanan , Fairness and Machine Learning , fairmlbook.org, 2019. http://www.fairmlbook.org

    work page 2019

  3. [3]

    Binns , On the apparent conflict between individual and group fairness , in Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency, 2020, pp

    R. Binns , On the apparent conflict between individual and group fairness , in Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency, 2020, pp. 514–524

    work page 2020

  4. [4]

    Borgonovo, A

    E. Borgonovo, A. Figalli, E. Plischke, and G. Savar ´e, Global sensitivity analysis via optimal transport, Management Science, (2024)

    work page 2024

  5. [5]

    Borgonovo, G

    E. Borgonovo, G. B. Hazen, V. R. R. Jose, and E. Plischke , Probabilistic sensitivity measures as information value , European Journal of Operational Research, 289 (2021), pp. 595–610

    work page 2021

  6. [6]

    Borgonovo, G

    E. Borgonovo, G. B. Hazen, and E. Plischke , A common rationale for global sensitivity measures and their estimation , Risk Analysis, 36 (2016), pp. 1871–1895

    work page 2016

  7. [7]

    M. C. Cohen, A. N. Elmachtoub, and X. Lei , Price discrimination with fairness constraints , Man- agement Science, 68 (2022), pp. 8536–8552

    work page 2022

  8. [8]

    Cˆot´e, M.-P

    O. Cˆot´e, M.-P. C ˆot´e, and A. Charpentier , A fair price to pay: exploiting causal graphs for fairness in insurance, Available at SSRN 4709243, (2024)

    work page 2024

Show all 41 references
  1. [9]

    Dutang and A

    C. Dutang and A. Charpentier , R package ‘casdatasets’ , 2020

    2020

  2. [10]

    Dwork, M

    C. Dwork, M. Hardt, T. Pitassi, O. Reingold, and R. Zemel , Fairness through awareness , in Proceedings of the 3rd innovations in theoretical computer science conference, 2012, pp. 214–226

    2012

  3. [11]

    FCA, General insurance pricing practices market study feedback to cp20/19 and final rules , Financial Conduct Authority (FCA) MS, (2021)

    2021

  4. [12]

    Fissler and S

    T. Fissler and S. M. Pesenti , Sensitivity measures based on scoring functions , European Journal of Operational Research, 307 (2023), pp. 1408–1423

    2023

  5. [13]

    E. W. J. Frees and F. Huang, The discriminating (pricing) actuary, North American Actuarial Journal, 27 (2023), pp. 2–24

    2023

  6. [14]

    Goldburd, A

    M. Goldburd, A. Khare, D. Tevet, and D. Guller , Generalized linear models for insurance rating, Casualty Actuarial Society, CAS Monographs Series, 5 (2016), p. 2016

    2016

  7. [15]

    Grari, C

    V. Grari, C. Arthur, L. Sylvain, and D. Marcin , A fair pricing model via adversarial learning , (2022), https://arxiv.org/abs/2202.12008

    2022 arXiv

  8. [16]

    Hardt, E

    M. Hardt, E. Price, and N. Srebro , Equality of opportunity in supervised learning , Advances in neural information processing systems, 29 (2016)

    2016

  9. [17]

    Hedden, On statistical criteria of algorithmic fairness , Philosophy and Public Affairs, 49 (2021)

    B. Hedden, On statistical criteria of algorithmic fairness , Philosophy and Public Affairs, 49 (2021)

    2021

  10. [18]

    Huang, J

    F. Huang, J. Shen, Y. Yang, and R. Zhao , Learning fair decisions with factor models: Applications to annuity pricing , 2025, https://arxiv.org/abs/2412.04663

    2025 arXiv

  11. [19]

    Huang and H

    F. Huang and H. Shimao, Welfare implications of fair and accountable insurance pricing , UNSW Busi- ness School Research Paper, (2025). (forthcoming)

    2025

  12. [20]

    Kleinberg, S

    J. Kleinberg, S. Mullainathan, and M. Raghavan , Inherent trade-offs in the fair determination of risk scores, (2016), https://arxiv.org/abs/1609.05807

    2016 arXiv

  13. [21]

    M. J. Kusner, J. R. Loftus, C. Russell, and R. Silva , Counterfactual fairness , in Proceedings of the 31st International Conference on Neural Information Processing Systems (NeurIPS), Curran Associates, Inc., 2017, pp. 4066–4076

    2017

  14. [22]

    Kusuoka , On law invariant coherent risk measures , Advances in Mathematical Economics, (2001), pp

    S. Kusuoka , On law invariant coherent risk measures , Advances in Mathematical Economics, (2001), pp. 83–95

    2001

  15. [23]

    Lindholm, R

    M. Lindholm, R. Richman, A. Tsanakas, and M. V. W ¨uthrich, Discrimination-free insurance pricing, ASTIN Bulletin: The Journal of the IAA, 52 (2022), pp. 55–89

    2022

  16. [24]

    Lindholm, A

    M. Lindholm, A. Tsanakas, and M. V. W ¨uthrich, Sensitivity-based measures of discrimination in insurance pricing, 2024

    2024

  17. [25]

    S. J. Mildenhall and J. A. Major , Pricing insurance risk: Theory and practice , John Wiley & Sons, 2022

    2022

  18. [26]

    https://www.naic.org/documents/committees c catf related price optimization white paper

    NAIC CAS Task Force , Casualty actuarial and statistical (c) task force price optimization white paper. https://www.naic.org/documents/committees c catf related price optimization white paper. pdf, 2015

    2015

  19. [27]

    S. M. Pesenti, S. Jaimungal, Y. F. Saporito, and R. S. Targino , Risk budgeting allocation for dynamic risk measures , Operations Research, (2024)

    2024

  20. [28]

    S. M. Pesenti, P. Millossovich, and A. Tsanakas , Cascade sensitivity measures, Risk Analysis, 41 (2021), pp. 2392–2414. 44 HUANG AND PESENTI

    2021

  21. [29]

    S. M. Pesenti, P. Millossovich, and A. Tsanakas , Differential sensitivity in discontinuous models , European Journal of Operational Research, 322 (2025), pp. 554–572

    2025

  22. [30]

    D. G. Pope and J. R. Sydnor, Implementing anti-discrimination policies in statistical profiling models , American Economic Journal: Economic Policy, 3 (2011), pp. 206–231

    2011

  23. [31]

    R¨uschendorf, Mathematical risk analysis, Springer Ser

    L. R¨uschendorf, Mathematical risk analysis, Springer Ser. Oper. Res. Financ. Eng. Springer, Heidelberg, (2013)

    2013

  24. [32]

    Singh, R

    S. Singh, R. Mehta, and M. Jaggi , Training individually fair ML models with sensitive subspace robustness, in International Conference on Learning Representations (ICLR), 2023

    2023

  25. [33]

    Tsanakas, To split or not to split: Capital allocation with convex risk measures , Insurance: Mathe- matics and Economics, 44 (2009), pp

    A. Tsanakas, To split or not to split: Capital allocation with convex risk measures , Insurance: Mathe- matics and Economics, 44 (2009), pp. 268–277

    2009

  26. [34]

    Tsanakas and P

    A. Tsanakas and P. Millossovich , Sensitivity analysis using risk measures , Risk Analysis, 36 (2016), pp. 30–48

    2016

  27. [35]

    W ang, R

    Q. W ang, R. W ang, and Y. Wei , Distortion riskmetrics on general spaces , ASTIN Bulletin: The Journal of the IAA, 50 (2020), pp. 827–851

    2020

  28. [36]

    W atson, F

    R. W atson, F. Mart´ınez-Plumed, and P. Flach , Fairness seen as global sensitivity analysis: A general approach for explainable fair classification , Machine Learning, 111 (2022), pp. 1995–2032

    2022

  29. [37]

    R. C. Williamson, A. K. Menon, G. Quon, W. R. Elbers, and A. J. Smola , Fairness risk mea- sures, in Proceedings of the 36th International Conference on Machine Learning, K. Chaudhuri and M. Sugiyama, eds., vol. 97 of Proceedings of Machine Learning Research, PMLR, 2019, pp. 6786–6797

    2019

  30. [38]

    Xin and F

    X. Xin and F. Huang, Antidiscrimination insurance pricing: Regulations, fairness criteria, and models , North American Actuarial Journal, 28 (2024), pp. 285–319

    2024

  31. [39]

    M. E. Yaari, The dual theory of choice under risk , Econometrica: Journal of the Econometric Society, (1987), pp. 95–115

    1987

  32. [40]

    Z. Yang, X. Fu, P. Gao, and Y.-J. Chen , Fairness regulation of prices in competitive markets , Man- ufacturing & Service Operations Management, 26 (2024), pp. 1897–1917

    2024

  33. [41]

    Zhang and H

    N. Zhang and H. Xu, Fairness of ratemaking for catastrophe insurance: Lessons from machine learning , Information Systems Research, 35 (2024), pp. 469–488

    2024

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.