Pith. sign in

REVIEW 3 major objections 4 minor 45 references

Generalized Orlicz premia

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read This paper proves that cash-additive generalized Orlicz premia are exactly L^p-quantiles, with expectiles as the convex case.

desk verdict Solid generalization of Orlicz premia with a repairable proof gap in Theorem 9 and an overbroad elicitability claim; the paper deserves refereeing. read the letter →

arxiv 2507.09181 v2 pith:DARRIRMA submitted 2025-07-12 q-fin.RM math.PRmath.STq-fin.MFstat.TH

Mücahit Aygün , Fabio Bellini , Roger J. A. Laeven This is my paper · ORCID
classification q-fin.RMmath.PRmath.STq-fin.MFstat.TH MSC 91B3046E30
keywords Orliczpremiacash-additivityLp-quantilesexpectilesgeometricconvexityCxLSpropertyelicitabilityreturnriskmeasures
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 extends the classical Orlicz premium principle by dropping convexity from the loss function, keeping only monotonicity, left-continuity, and a normalization around the unit level, and shows that the geometric mean, ordinary quantiles, expectiles, and $L^p$-quantiles are all generalized Orlicz premia. Its central theorem states that if such a premium is cash-additive and the loss function is finite, continuous, and strictly increasing, the loss function must be the asymmetric power form $1+a(x-1)^p_+ - b(x-1)^p_-$, so the premium is an $L^p$-quantile; adding convexity or concavity forces $p=1$, giving an expectile. This replaces the classical collapse-to-the-mean result for convex Orlicz premia with a collapse-to-$L^p$-quantiles result. The paper also characterizes geometric convexity of the premium through GA-convexity of the loss function, derives multiplicative dual representations, and shows that law-invariant geometrically convex return risk measures with convex level sets under mixtures are exactly Orlicz premia.

What carries the argument

The load-bearing object is the generalized Orlicz premium $H_\Phi(X)=\inf\{k>0: \mathbb{E}[\Phi(X/k)]\le 1\}$ on nonnegative bounded losses, with $\Phi$ an Orlicz function: nondecreasing, left-continuous, $\Phi\le1$ below 1 and $\Phi>1$ above 1, but not necessarily convex. Theorem 9 evaluates the premium on two-point claims $X=x_1 1_A+x_2 1_{A^c}$ with $x_1<1<x_2$, uses cash-additivity and positive homogeneity to obtain $H_\Phi(cX-c+1)=1$, and converts this into the identity $p\Phi(c(x_1-1)+1)+(1-p)\Phi(c(x_2-1)+1)=1$, where $p=\mathbb{P}(A)$. Solving the resulting multiplicative Pexider functional equation yields the asymmetric power tails $f(u)=-a(-u)^p$ and $g(v)=b v^p$, which is precisely the structure that defines an $L^p$-quantile.

What would settle it

A concrete way to test Theorem 9 is to take any finite, continuous, strictly increasing Orlicz function $\Phi$ not of the form $1+a(x-1)^p_+ - b(x-1)^p_-$, and check whether its premium is cash-additive on two-point claims for scaling factors beyond the range where $cX-c+1$ stays nonnegative; if the cash-additivity identity holds there without an extension of $\Phi$ to negative arguments, the theorem or its proof would need revision.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that cash-additivity does not force an Orlicz premium down to the expectation unless convexity of the loss function is also assumed. If $\Phi$ is a finite, continuous, strictly increasing Orlicz function, then $H_\Phi(X+h)=H_\Phi(X)+h$ forces $\Phi(x)=1+a(x-1)^p_+ - b(x-1)^p_-$ with $a,b>0$ and $p\ge0$, and this $\Phi$ is exactly the loss function whose premium is an $L^p$-quantile. Restoring convexity or concavity gives $p=1$, the expectile case, and geometric convexity has the same effect. The paper reads this as replacing 'collapse to the mean' by 'collapse to $L^p$-quantiles', with the classical mean result appearing as the convex special case.

Load-bearing premise

The characterization assumes that the shifted claim $cX-c+1$ can be fed into the premium for every $c\ge0$, although this claim becomes negative when $c$ is large and $X$ is below 1, and the premium is only defined on nonnegative bounded losses; no extension to negative values is stated.

Editorial extensions

If this is right

  • Cash-additive generalized Orlicz premia are exactly $L^p$-quantiles, so no other Orlicz premium in the generalized class can be cash-additive.
  • Convex or concave cash-additive generalized Orlicz premia are exactly expectiles, making expectiles the unique convex cash-additive Orlicz premia.
  • A generalized Orlicz premium is geometrically convex exactly when its Orlicz function is GA-convex, and cash-additive geometrically convex premia are again expectiles.
  • Geometrically convex Orlicz premia admit a dual representation as a multiplicatively penalized supremum of geometric means, the multiplicative analogue of the arithmetic dual representation.
  • Among law-invariant, positively homogeneous, monotone, normalized return risk measures, the CxLS property characterizes Orlicz premia under geometric convexity or convexity, connecting the class to elicitability.

Reading between the lines

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

  • If the theorem extends to signed losses, cash-additivity may characterize a broader family of two-sided power Orlicz functions $1+a(x-1)^p_+ - b(x-1)^q_-$; the paper's own Example 10 suggests that cash-additivity would then force $p=q$.
  • One could read cash-additivity as a symmetry axiom that selects asymmetric power loss functions among all monotone normalized loss functions; testing related axioms, such as subadditivity or comonotonic additivity, against the generalized Orlicz family may yield analogous characterizations.
  • A multiplicative analogue of Theorem 9, replacing cash-additivity by a translation property in log returns, would likely axiomatize geometric $L^p$-quantiles and geometric expectiles, paralleling the paper's geometric-expectile example.
  • The CxLS and elicitability result suggests that every law-invariant geometrically convex Orlicz premium can be elicited by some expected loss; identifying explicit scoring functions for the whole class is a natural next step that the paper leaves to future work.
Share X Bluesky LinkedIn Reddit HN

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 introduces a generalized class of Orlicz premia obtained by dropping convexity of the loss function Phi and allowing extended values, with the domain restricted to L^8_+. It shows that the basic properties of classical Orlicz premia (monotonicity, positive homogeneity, normalization, law-invariance) persist, and it proves that cash-additivity forces Phi to have the form 1 + a(x-1)^p_+ - b(x-1)^p_- (Theorem 9), which yields L^p-quantiles and, under additional convexity or concavity, expectiles. The paper then studies geometric convexity of such premia, provides dual representations in both the convex and geometrically convex cases, and gives a CxLS-based axiomatization of geometrically convex and convex Orlicz premia. The central claim is the 'collapse-to-L^p-quantiles' result in Theorem 9, which is meant to replace the classical collapse-to-the-mean for convex Orlicz premia.

Significance. If Theorem 9 is correct, the paper delivers a genuine unification: quantiles, expectiles, and L^p-quantiles all appear as generalized Orlicz premia, and cash-additivity is shown to single out exactly the L^p-quantile family. The paper is rich in worked examples, the geometric convexity perspective is original and interesting, and the duality results (Theorems 18 and 21) together with the CxLS characterization (Theorem 28) are valuable contributions. The appendix is self-contained for most auxiliary results. However, the proof of the central Theorem 9 contains a domain gap: cash-additivity is applied to random variables that take negative values, outside the stated domain L^8_+ and outside the domain of definition of Phi. As written, the full-half-line Pexider equation is not justified, so the main claim is not established. The gap appears potentially repairable, but it is load-bearing and requires a substantive revision of the proof.

major comments (3)
  1. [Appendix A, Proof of Theorem 9] The identity H(cX-c+1)=1 is applied for every c>=0. Since X takes the value x1<1 on the event A, the random variable cX-c+1 equals 1+c(x1-1) on A, which is negative for c>1/(1-x1). The domain of H is L^8_+ and Phi is defined only on [0,infinity), so the cash-additivity step and the resulting equation E[Phi(1+c(X-1))]=1 are justified only for c in [0,1/(1-x1)]. Consequently the multiplicative Pexider functional equation is derived only on a bounded c-interval depending on u, not on the full half-line, and the power form of Phi does not follow as written. This is the central claim of the paper, so the proof must be repaired, for example by a localization argument that still yields the power form on the full half-line or by explicitly extending H to all bounded random variables.
  2. [Section 2 and Theorem 9] Cash-additivity is stated informally as H(X+h)=H(X)+h for all h in R, but H is defined only on L^8_+. This is not merely a notational imprecision: the proof of Theorem 9 uses exactly this property for arguments outside L^8_+. The paper should state the precise domain of the cash-additivity axiom, either restricting it to X, X+h in L^8_+, or extending H to all of L^8 and verifying that the extension preserves the other assumptions. Without such a clarification, the definition does not justify the steps in the proof.
  3. [Theorem 9 and Example 3] The sentence 'letting alpha = a/(a+b) leads to Examples 3, 4 and 5' is inaccurate for Example 3, because the quantile Orlicz function is a step function that is neither continuous nor strictly increasing, and therefore does not satisfy the hypotheses of Theorem 9. Since ordinary quantiles are cash-additive, the theorem as stated does not actually cover the quantile case; the p=0 case in the power form is not attained under the strict-monotonicity assumption. The text should either state that the quantile case arises only in a limiting or regularity-relaxing sense, or the theorem should be extended under weaker assumptions on Phi.
minor comments (4)
  1. [Section 4] There is a typo 'artihmetic' that should read 'arithmetic'.
  2. [Proof of Theorem 9] The letter p is used both for the probability P(A) and for the exponent in the Pexider solution; consider using a different symbol, for instance q, for the exponent to avoid confusion.
  3. [Section 6] The concluding statement that generalized Orlicz premia are 'the broadest class of elicitable, positively homogeneous and monotonic premium principles' overstates the formal results: Theorem 28 and Corollary 29 require law-invariance, geometric convexity (or convexity), and the CxLS property, and CxLS is only necessary for elicitability, not sufficient.
  4. [Example 5] The phrase 'L^p+1-quantile' may confuse readers; a parenthetical explanation that p is the exponent in the defining inequality and corresponds to the derivative order of the underlying L^{p+1} minimization would be helpful.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the cash-additivity-to-L^p-quantile theorem is derived from the Pexider equation; self-citations are auxiliary tools, not load-bearing inputs.

full rationale

The central derivation chain is self-contained. Theorem 9 starts from cash-additivity, namely H_Phi(cX-c+1)=1, combines it with Proposition 8(e) to obtain E[Phi(1+c(X-1))]=1, and then solves the resulting multiplicative Pexider equation with the classical solution f(u)=-a(-u)^p and g(v)=b v^p. The definition of the Orlicz premium and the cash-additivity assumption do not already contain the L^p-quantile form; the conclusion is genuinely derived rather than assumed. The examples in Section 2 are verification checks, not fitted predictions. Proposition 14 is proved directly from the GA-convexity inequality, and Theorem 18 uses external duality results from [33] and [22]. The self-citations to [9] and [3] are used as structural tools: Lemma 2 of [9] converts a return risk measure into a monetary risk measure via log and exp, and [3] is referenced only as related work on general geometric convexity. Theorem 28 is anchored in the external theorem of Delbaen et al. (2016) [19], not in a self-citation chain. The only visible weakness is a domain-restriction issue in the proof of Theorem 9, where cash-additivity is applied to cX-c+1 for all c>=0 even though 1+c(X-1) can become negative on the event {X=x1} for large c; this is a correctness gap concerning the validity of the Pexider equation on the full half-line, not a circularity. No fitted parameter is renamed as a prediction, and no result is defined in terms of its own conclusion. Therefore the appropriate finding is no significant circularity, with a low score reflecting only the presence of minor, non-load-bearing self-citations.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

The paper is a pure mathematics work; no data are fitted and no new physical or probabilistic entities are postulated. The central claims rest on standard functional-analytic results and two external characterization theorems from previous literature that are used without reproof.

assumptions (6)
  • domain assumption Proposition 30 from Delbaen, Bellini, Bignozzi, Ziegel (2016): a convex, law-invariant, monotone, cash-additive risk measure with CxLS has a representation via a convex phi.
    Invoked as a black box in the proof of Theorem 28 to construct the Orlicz function Phi.
  • domain assumption Lemma 2 from Bellini, Laeven, Rosazza Gianin (2018): the log-transform maps return risk measures to monetary risk measures, preserving Fatou/Lebesgue and CxLS properties.
    Used in Theorems 21 and 28 to convert geometric convexity into ordinary convexity.
  • standard math Standard dual representation for convex monetary risk measures with the Fatou property (e.g., Foellmer and Schied).
    Foundation for Theorem 18 and Theorem 21.
  • domain assumption The probability space is non-atomic (stated at the start of Section 3).
    Needed for the quantile-function representations of law-invariant functionals used in the dual representations.
  • standard math Pexider functional equation solution from Aczel (1966), Theorem 4 in Section 3.1.
    Used to solve the functional equation in Theorem 9, yielding power functions.
  • standard math Donsker-Varadhan variational formula for relative entropy.
    Used in the proof of Proposition 24 to link arithmetic and geometric dual representations.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Generalized Orlicz premia." pith.science (2026). https://pith.science/paper/DARRIRMA

@misc{pith2026250709181,
  author       = {Pith},
  title        = {Pith review of: Generalized Orlicz premia},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DARRIRMA}},
  note         = {Machine review of arXiv:2507.09181}
}
abstract

We introduce a generalized class of Orlicz premia based on possibly nonconvex loss functions, extending the classical framework of Haezendonck and Goovaerts (1982). Without the usual convexity requirement on the loss function $\Phi$, the Orlicz framework naturally encompasses quantiles, expectiles and $L^p$-quantiles while preserving the fundamental properties of Orlicz premia. We show that within this framework cash-additivity axiomatizes $L^p$-quantiles, generalizing the classical `collapse-to-the-mean' result for cash-additive convex Orlicz premia into a `collapse-to-$L^p$-quantiles' result, with expectiles as a special case. We focus on two natural classes of nonconvex loss functions: concave-convex Orlicz functions, which mimic the idea of S-shaped value functions in prospect theory, and GA-convex Orlicz functions, which can be described in terms of comparative convexity with respect to a logarithmic reference, and for which the corresponding Orlicz premium is geometrically convex. Finally, we show that a suitable subclass of generalized Orlicz premia coincides with the class of law-invariant, monotone, positive, positively homogeneous, normalized functionals that are weakly lower semicontinuous, continuous from above, and whose level sets are convex with respect to mixtures (the so-called CxLS property).

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 42 canonical work pages

  1. [1]

    Acz´el, J. (1966). Lectures on Functional Equations and Their Applica- tions. Academic Press, New York

    work page 1966

  2. [2]

    D., Border, K

    Aliprantis, C. D., Border, K. C.(2006). Infinite Dimensional Analysis. Third edition, Springer Verlag, Berlin

    work page 2006

  3. [3]

    Ayg¨un, M., Bellini, F., Laeven, R. J. A. (2024). On geometrically convex risk measures. Available at: https://doi.org/10.48550/arXiv. 2403.06188

    work page Pith review arXiv doi:10.48550/arxiv.2403.06188 2024

  4. [4]

    Bauer, D., Zanjani, G. (2016). The marginal cost of risk, risk measures, and capital allocation. Management Science 62(5), 1431-1457

    work page 2016

  5. [5]

    Bellini, F., Bignozzi, V. (2015). On elicitable risk measures. Quantita- tive Finance 15(5), 725-733

    work page 2015

  6. [6]

    Bellini, F., Rosazza Gianin, E. (2008). On Haezendonck risk measures. Journal of Banking and Finance 32(6), 986–994

    work page 2008

  7. [7]

    Bellini, F., Rosazza Gianin, E. (2012). Haezendonck–Goovaerts risk measures and Orlicz quantiles. Insurance: Mathematics and Economics 51, 107-114

    work page 2012

  8. [8]

    Bellini, F., Klar, B., M ¨uller, A., Rosazza Gianin, E. (2014). Gen- eralized quantiles as risk measures.Insurance: Mathematics and Economics 54, 41-48

    work page 2014

Show all 45 references
  1. [9]

    Bellini, F., Laeven, R. J. A., Rosazza Gianin, E. (2018). Robust return risk measures. Mathematics and Financial Economics 12(1), 5-32. 22

    2018

  2. [10]

    Bellini, F., Laeven, R. J. A., Rosazza Gianin, E. (2021). Dynamic robust Orlicz premia and Haezendonck-Goovaerts risk measures. European Journal of Operational Research 291(2), 438-446

    2021

  3. [11]

    Ben-Tal, A., Teboulle, M. (1986). Expected utility, penalty functions, and duality in stochastic nonlinear programming. Management Science 32(11), 1445-1466

    1986

  4. [12]

    Ben-Tal, A., Teboulle, M. (2007). An old-new concept of convex risk measures: The optimized certainty equivalent. Mathematical Finance 17(3), 449-476

    2007

  5. [13]

    Biagini, S., Frittelli, M. (2008). A unified framework for utility maxi- mization problems: An Orlicz space approach. The Annals of Applied Prob- ability 18(3), 929-966

    2008

  6. [14]

    Canna, G., Centrone, F., Rosazza Gianin, E. (2021). Haezendonck- Goovaerts capital allocation rules. Insurance: Mathematics and Economics 101, Part B, 173-185

    2021

  7. [15]

    Chen, Z. (1996). Conditional Lp-quantiles and their application to testing of symmetry in non-parametric regression. Statistics & Probability Letters 29, 107-115

    1996

  8. [16]

    Cheridito, P., Li, T. (2008). Dual characterization of properties of risk measures on Orlicz hearts. Mathematics and Financial Economics 2(29), 29-55

    2008

  9. [17]

    Cheridito, P., Li, T. (2009). Risk measures on Orlicz Hearts. Mathe- matical Finance 19(2), 189-214

    2009

  10. [18]

    Delbaen, F. (2012). Monetary Utility Functions. Osaka University Press, Osaka

    2012

  11. [19]

    Delbaen, F., Bellini, F., Bignozzi, V., Ziegel, J. F. (2016). Risk measures with the CxLS property. Finance and Stochastics 20(2), 433-453

    2016

  12. [20]

    D., V aradhan, S

    Donsker, M. D., V aradhan, S. R. S.(1983). Asymptotic evaluation of certain Markov process expectations for large time. IV. Communications on Pure and Applied Mathematics 36(2), 183-212

    1983

  13. [21]

    Fissler, T., Ziegel, J. F. (2016). Higher order elicitability and Osband’s principle. The Annals of Statistics 44(4), 1680-1707

    2016

  14. [22]

    F¨ollmer, H., Schied, A. (2016). Stochastic Finance. Fourth edition, De Gruyter, Berlin

    2016

  15. [23]

    (2002) Convex measures of risk and trading constraints

    F¨ollmer, H., Schied, A. (2002) Convex measures of risk and trading constraints. Finance & Stochastics 6, 429-447

    2002

  16. [24]

    (2002) Putting order in risk mea- sures

    Frittelli, M., Rosazza Gianin, E. (2002) Putting order in risk mea- sures. Journal of Banking & Finance 26 (7), 1473-1486

    2002

  17. [25]

    Gneiting, T. (2011). Making and evaluating point forecasts. Journal of the American Statistical Association 106(494), 746-762. 23

    2011

  18. [26]

    J., De Vylder, F

    Goovaerts, M. J., De Vylder, F. E. C., Haezendonck, J. (1984). Insurance Premiums. North-Holland Publishing, Amsterdam

    1984

  19. [27]

    J., Kaas, R., Dhaene, J., Tang, Q

    Goovaerts, M. J., Kaas, R., Dhaene, J., Tang, Q. (2004). Some new classes of consistent risk measures. Insurance: Mathematics and Economics 34, 505-516

    2004

  20. [28]

    Haezendonck, J., Goovaerts, M. J. (1982). A new premium calcu- lation principle based on Orlicz norms. Insurance: Mathematics and Eco- nomics 1(1), 41-53

    1982

  21. [29]

    Laeven, R. J. A., Rosazza Gianin, E. (2022). Quasi-logconvex mea- sures of risk. Available at: https://arxiv.org/abs/2208.07694

    2022 arXiv

  22. [30]

    Laeven, R. J. A., Rosazza Gianin, E., Zullino, M. (2025). Dynamic return and star-shaped risk measures via BSDEs. Finance and Stochastics, in press

    2025

  23. [31]

    Laeven, R. J. A., Rosazza Gianin, E., Zullino, M. (2024a). Law- invariant return and star-shaped risk measures. Insurance: Mathematics and Economics 117, 140-153

    2024

  24. [32]

    Laeven, R. J. A., Rosazza Gianin, E., Zullino, M. (2024b). Geomet- ric BSDEs. Available at: https://arxiv.org/abs/2405.09260

    2024 arXiv

  25. [33]

    Laeven, R. J. A., Stadje, M. (2013). Entropy coherent and entropy convex measures of risk. Mathematics of Operations Research 38(2), 265- 293

    2013

  26. [34]

    Laeven, R. J. A., Stadje, M. (2014). Robust portfolio choice and indif- ference valuation. Mathematics of Operations Research 39(4), 1109-1141

    2014

  27. [35]

    Mohammed, N., Furman, E., Su, J. (2021). Can a regulatory risk mea- sure induce profit-maximizing risk capital allocations? The case of condi- tional tail expectation. Insurance: Mathematics and Economics 101, Part B, 425-436

    2021

  28. [36]

    K., Powell, J

    Newey, W. K., Powell, J. L. (1987). Asymmetric least squares estima- tion and testing. Econometrica 55(4), 819-847

    1987

  29. [37]

    Niculescu, C. P. (2000). Convexity according to the geometric mean. Mathematical Inequalities & Applications 3(2), 155-167

    2000

  30. [38]

    P., Persson L.-E

    Niculescu, C. P., Persson L.-E. (2004). Convex Functions and Their Applications. Springer, New York

    2004

  31. [39]

    Osband, K. H. (1985). Providing Incentives for Better Cost Forecasting. PhD Thesis, University of California, Berkeley

    1985

  32. [40]

    M., Ren, Z

    Rao, M. M., Ren, Z. D. (1991). Theory of Orlicz Spaces. Marcel Dekker, New York

    1991

  33. [41]

    T., Wets, R

    Rockafellar, R. T., Wets, R. J. B. (2009). Variational Analysis. Third edition, Springer, Berlin

    2009

  34. [42]

    Risk functionals with convex level sets

    W ang, R., Wei, Y.(2020). Risk functionals with convex level sets. Math- ematical Finance 30, 1337-1367. 24

    2020

  35. [43]

    Weber, S. (2006). Distribution-invariant risk measures, information, and dynamic consistency. Mathematical Finance 16(2), 419-441

    2006

  36. [44]

    Wu, Q., Xu, H. (2022). Robust distorted Orlicz premium: Modelling, computational scheme and applications. Available at: https://ssrn.com/ abstract=4093580 or http://dx.doi.org/10.2139/ssrn.4093580

    2022 doi

  37. [45]

    Ziegel, J. F. (2016). Coherence and elicitability. Mathematical Finance 26(4), 901-918. 25

    2016

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.