betaFeedback

WorksMathematics

A characterization and closure property of Thorin's class

Abstract

Thorin's class coincides with the smallest class of probability laws on R\mathbb{R} that contains all gamma laws and their reflections, while being closed under convolution and weak convergence. In particular, all Thorin laws are self-decomposable. This paper proves an alternative characterization of Thorin's class in terms of what we call the hyperbolic Bochner-Bernstein property. This property corresponds to the notion of hyperbolically completely monotone Laplace transforms on the positive half line, but extended to the real-valued setting. Using the new characterization, it is shown that Thorin's class is closed under multiplication with independent random variables admitting hyperbolically completely monotone densities, which partly settles an open question raised by Bondesson and allows for proving that a large class of possibly skewed real-valued distributions is in Thorin's class. As particular examples, we apply the closure property to show that Johnson's S\textsubscript{U} distribution and the Pearson type IV distribution are in Thorin's class for all parameter choices.

Introduction

Thorin’s class of probability laws introduced in Thorin [26] extends the class of generalized gamma convolutions to the real line. The generalized gamma convolutions were first introduced in Thorin [24, 25] to study infinite divisibility of some classical probability laws and a comprehensive modern treatment of these and Thorin’s class can be found in Bondesson’s monograph; Bondesson [7]. Thorin’s class can be characterized as the smallest class that contains all gamma laws on (0,∞)(0, \infty) and their reflections, while being closed under convolution and weak convergence; multivariate extensions based on this definition are studied in Barndorff-Nielsen et al. [3] and Torricelli [27]. In the following, we shall denote Thorin’s class by T(R)\mathcal{T}(\mathbb{R}) and we note that it is closed under weak limits and affine maps, and contains all Gaussian and degenerate laws. Every law in T(R)\mathcal{T}(\mathbb{R}) is self-decomposable and thus corresponds to the stationary law of a Lévy-driven Ornstein-Uhlenbeck process.

The main result of this paper is a characterization of T(R)\mathcal{T}(\mathbb{R}) in terms of a property of the characteristic function of the laws in T(R)\mathcal{T}(\mathbb{R}), which we term the hyperbolic Bochner-Bernstein property. This is the real-line analogue of hyperbolically completely monotone Laplace transforms cf. Remark 3.1 below, which characterizes the class of generalized gamma convolutions cf. Bondesson [6].

Definition 1.1. A probability law η\eta on R\mathbb{R} is said to be hyperbolically Bochner-Bernstein (HBB) with Bochner-Bernstein measure μ\mu if there is a finite positive measure μ\mu on R×R\mathbb{R} \times\mathbb{R} such that, for all u,w>0u,w>0, ∬∣ei(u−w)x−uwy∣μ(dx,dy)<∞\iint\lvert e^{\mathrm{i}(u-w)x-uwy}\rvert\mu(\mathrm{d}x,\mathrm{d}y)<\infty and the Fourier transform of η\eta, Fη\mathcal{F}_{\eta}, satisfies

Fη(u)Fη(−w)=∫R∫Rei(u−w)x−uwy μ(dx,dy).(1)\mathcal{F}_{\eta}(u)\mathcal{F}_{\eta}(-w)=\int_{\mathbb{R}}\int_{\mathbb{R}} e^{\mathrm{i}(u-w)x-uwy}\,\mu(\mathrm{d}x,\mathrm{d}y). \tag*{(1)}

For a random variable XX with an HBB law, we write X∈HBBX \in\mathcal{H}_{BB}.

The main result of the paper is then the following theorem.

Theorem 1.1. A law belongs to T(R)\mathcal{T}(\mathbb{R}) if and only if it is hyperbolically Bochner-Bernstein.

The proof of Theorem 1.1 relies on similar methods as those in Chapter 5 and 7 of Bondesson [7], but require some additional care since we must treat the complex-valued Fourier transform as opposed to the more convenient Laplace transform in the case of positive random variables. The hyperbolic Bochner-Bernstein property can be equivalently stated as existence of a particular coupling (Z,Y)(Z,Y), where Z=dXZ \mathrel{\overset{d}{=}} X and YY carries information about the symmetric part of ZZ. In particular, it is shown in Proposition 3.2 that YY is a generalized gamma convolution and that the symmetrization of XX has the law of the random variable 2YN\sqrt{2Y}N, where NN is standard Gaussian.

Having established Theorem 1.1, the following corollary follows from Theorem 3.1 below, which partly settles the first open problem posed in Section 9 of Bondesson [8].

Corollary 1.1. Let XX be a random variable with law in T(R)\mathcal{T}(\mathbb{R}) and let YY be an independent and strictly positive random variable that admits a hyperbolically completely monotone density. Then the laws of YXYX and X/YX/Y are also in T(R)\mathcal{T}(\mathbb{R}).

The class of probability laws with hyperbolically completely monotone densities is a strict subclass of those with hyperbolically completely monotone Laplace transforms and the closure property of Corollary 1.1 is well-known in the case of generalized gamma convolutions where the product part of the result remains valid with YY a generalized gamma convolution due to Theorem 1 of Bondesson [8]. For symmetric Thorin laws, the product part also holds with multipliers Y\sqrt{Y} with YY a generalized gamma convolution cf. Theorem 2 of Bondesson [8]. Whether a general version holds for all Thorin laws is posed as an open problem in Section 9 of Bondesson [8] and Corollary 1.1 then settles this problem for a large class of multipliers. The departure from the symmetric setting requires a different strategy than the Gaussian mixture representations typically used in, e.g., Behme et al. [4], Bondesson [8], and Torricelli [27].

As an application of the developed theory, we apply the closure property of Corollary 1.1 to establish that the law of a certain class of exponential functionals of Lévy processes that typically arise in insurance risk modelling is in T(R)\mathcal{T}(\mathbb{R}). This is the content of § 4 and as particular examples of this result, we show that Johnson’s SU\mathrm{S}_{\mathrm{U}} distribution as introduced in Johnson [13] and the Pearson type IV distribution belong to T(R)\mathcal{T}(\mathbb{R}). Both of these distributions are frequently applied as flexible models for skewed and heavy-tailed real-valued data, but their Thorin class membership has, to the best of our knowledge, never been shown previously. In fact, infinite divisibility or self-decomposability of Johnson’s SU\mathrm{S}_{\mathrm{U}} distribution appears to be a novel claim, whereas self-decomposability of the Pearson type IV distribution can be deduced from known results (see Example 4.2). In § 4.2, it is argued that the Gaussian mixture approach of the symmetric case fails for these distributions, thus necessitating an alternative approach, which the hyperbolic Bochner-Bernstein property provides.

Preliminaries and classes of distributions

In this section, we briefly recap the various classes of probability laws that will be used throughout and some of their most important properties. Throughout, we fix a probability space (Ω,F,P)(\Omega,\mathcal{F},\mathbb{P}), on which all random variables are defined. If a random variable XX has law η=P∘X−1\eta=\mathbb{P}\circ X^{-1}, then we write X∼ηX\sim\eta and if S\mathcal{S} denotes a class of probability laws, then we write X∈SX\in\mathcal{S} if η∈S\eta\in\mathcal{S}. For a probability law η\eta, we denote its Fourier transform by Fη(u)=∫eiuxη(dx)\mathcal{F}_{\eta}(u)=\int e^{\mathrm{i}ux}\eta(\mathrm{d}x) for u∈Ru\in\mathbb{R} and its Laplace transform by Lη(u)=∫e−uxη(dx)\mathcal{L}_{\eta}(u)=\int e^{-ux}\eta(\mathrm{d}x) for u∈(0,∞)u\in(0,\infty). When X∼ηX\sim\eta, we often use the notation φX(u)\varphi_X(u) for the characteristic function of XX, i.e., φX(u)=Fη(u)\varphi_X(u)=\mathcal{F}_{\eta}(u). If it is obvious from the context that φX\varphi_X pertains to XX, we shall sometimes simply use φ(u)\varphi(u) for φX(u)\varphi_X(u) to avoid excessive notation.

We shall be concerned with infinitely divisible probability laws, which are all probability laws η\eta with Fourier transform of Lévy-Khintchine type [see 21]

Fη(u)=exp⁡(iuβ−σ2u22+∫R(eiux−1−iux1(−1,1)(x))ν(dx)),(2)\mathcal{F}_{\eta}(u)=\exp\left(\mathrm{i}u\beta-\frac{\sigma^2u^2}{2}+\int_{\mathbb{R}}\left(e^{\mathrm{i}ux}-1-\mathrm{i}ux1_{(-1,1)}(x)\right)\nu(\mathrm{d}x)\right), \tag*{(2)}

where ν\nu is the Lévy measure and satisfies ∫(1∧x2)ν(dx)<∞\int(1\wedge x^2)\nu(\mathrm{d}x)<\infty. If the Lévy measure is of the form ν(dx)=g(x)∣x∣dx\nu(\mathrm{d}x)=\frac{g(x)}{|x|}\mathrm{d}x where the function g(x)g(x) is non-negative for all x∈Rx\in\mathbb{R}, non-decreasing on (−∞,0)(-\infty,0), and non-increasing on (0,∞)(0,\infty), then we say that η\eta is self-decomposable. The self-decomposable distributions are unimodal and correspond to stationary solutions of Lévy-driven Ornstein-Uhlenbeck equations cf. [14], [22], and [29].

Definition 2.1 (Generalized gamma convolution). A probability law η\eta on [0,∞)[0,\infty) is a generalized gamma convolution (GGC) if its Laplace transform has the form

Lη(u)=exp⁡(−au−∫(0,∞)log⁡(1+uy)U(dy)),u≥0,(3)\mathcal{L}_{\eta}(u)=\exp\left(-au-\int_{(0,\infty)}\log\left(1+\frac{u}{y}\right)U(\mathrm{d}y)\right),\qquad u\geq0, \tag*{(3)}

where a≥0a\geq0 and UU is a non-negative measure, termed the Thorin measure, satisfying

∫(0,1]∣log⁡(y)∣U(dy)<∞,∫(1,∞)y−1U(dy)<∞.\int_{(0,1]}|\log(y)|U(\mathrm{d}y)<\infty,\qquad\int_{(1,\infty)}y^{-1}U(\mathrm{d}y)<\infty.

For a random variable XX with a GGC law, we write X∈T(R+)X\in\mathcal{T}(\mathbb{R}_+).

Definition 2.2 (Hyperbolic complete monotonicity). A function ff on (0,∞)(0,\infty) is hyperbolically completely monotone (HCM) if for every u>0u>0, the function v↦f(uv)f(u/v)v\mapsto f(uv)f(u/v) is completely monotone as a function of w=v+v−1∈[2,∞)w=v+v^{-1}\in[2,\infty). For a random variable XX with an HCM density, we write X∈HCMX\in\mathcal{H}_{CM}.

The GGC class was introduced in Thorin [24, 25] and can equivalently be characterized as the closure under weak convergence of the set of all finite convolutions of gamma laws. It is closed under convolution and scaling – hence affine maps with positive coefficients – and taking U≡0U\equiv0 in (3) shows that every Dirac distribution δa\delta_a with a≥0a\geq0 is a GGC. Every GGC is infinitely divisible and self-decomposable. The term “hyperbolic complete monotonicity” was first labelled in Bondesson [7], although the property has been implicitly used in previous works on GGC’s, including those of Thorin. Every HCM density corresponds to a GGC law, but HCM\mathcal{H}_{CM} is a strict subset of the non-degenerate laws in T(R+)\mathcal{T}(\mathbb{R}_+). The GGC laws can, however, be characterized as the laws that admit HCM Laplace transforms; this was in fact the route taken in Thorin [25] to show that the lognormal distribution is infinitely divisible. Furthermore, it is shown in Bondesson [7] and Steutel and van Harn [23] that if X∈T(R+)X \in\mathcal{T}(\mathbb{R}_{+}) and Y∈HCMY \in\mathcal{H}_{CM} are independent, then XY∈T(R+)XY \in\mathcal{T}(\mathbb{R}_{+}), which is upgraded in Theorem 1 of Bondesson [8] to the fact that for independent X,Y∈T(R+)X,Y \in\mathcal{T}(\mathbb{R}_{+}) then XY∈T(R+)XY \in\mathcal{T}(\mathbb{R}_{+}).

Definition 2.3 (Thorin’s class T(R)\mathcal{T}(\mathbb{R})). A probability law η\eta on R\mathbb{R} is in Thorin’s class, T(R)\mathcal{T}(\mathbb{R}), if its Fourier transform has the form

Fη(u)=exp⁡(iuβ−σ2u22+∫R∖{0}[log⁡(xx−iu)−iux1+x2]U(dx)),(4)\mathcal{F}_{\eta}(u)=\exp\left(iu\beta-\frac{\sigma^{2}u^{2}}{2}+\int_{\mathbb{R}\setminus\{0\}}\left[\log\left(\frac{x}{x-iu}\right)-\frac{iux}{1+x^{2}}\right]U(dx)\right), \tag*{(4)}

where β∈R\beta\in\mathbb{R}, σ≥0\sigma\ge0, and UU is a non-negative measure on R∖{0}\mathbb{R}\setminus\{0\}, termed the Thorin measure, such that

∫R11+x2U(dx)<∞,and∫{∣x∣≤1}∣log⁡(x2)∣U(dx)<∞.(5)\int_{\mathbb{R}}\frac{1}{1+x^{2}}U(dx)<\infty,\qquad\text{and}\qquad\int_{\{|x|\le1\}}|\log(x^{2})|U(dx)<\infty. \tag*{(5)}

The equivalence of Definition 2.3 with the weak closure of finite gamma convolutions and their reflections is proven in Bondesson [7] and Thorin [26]. Comparing the representation (4) to the classical Lévy-Khintchine representation (2), we see that iuβiu\beta represents the drift plus truncation, σ2u22\frac{\sigma^{2}u^{2}}{2} the Gaussian part, and the integral against UU the jump structure. It is shown in Bondesson [7] that any X∈T(R)X \in\mathcal{T}(\mathbb{R}) is self-decomposable. The Thorin measure UU can be decomposed as

∫Rf(x)U(dx)=∫0∞f(x)U+(dx)+∫0∞f(−x)U−(dx),\int_{\mathbb{R}}f(x)U(dx)=\int_{0}^{\infty}f(x)U_{+}(dx)+\int_{0}^{\infty}f(-x)U_{-}(dx),

where ff is any positive measurable function and

U+(B)=U(B),U−(B)=U(−B),B⊆(0,∞).U_{+}(B)=U(B),\qquad U_{-}(B)=U(-B),\qquad B\subseteq(0,\infty).

References

  1. [1]Alili, L., Dufresne, D., & Yor, M. (1997). Sur l’identité de Bougerol pour les fonctionnelles exponentielles du mouvement Brownien avec drift. In M. Yor (Ed.), Exponential functionals and principal values related to Brownian motion. a collection of research papers (pp. 3–14). Biblioteca de la Revista Mathematica, Iberoamericana.
  2. [2]Barndorff-Nielsen, O. E., & Halgreen, C. (1977). Infinite divisibility of the hyperbolic and generalized inverse Gaussian distributions. Zeitschrift für Wahrscheinlichkeitstheorie und Verwandte Gebiete, (38), 309–311. https://doi.org/10.1007/BF00533162
  3. [3]Barndorff-Nielsen, O. E., Maejima, M., & Sato, K.-I. (2006). Some classes of multivariate infinitely divisible distributions admitting stochastic integral representations. Bernoulli : official journal of the Bernoulli Society for Mathematical Statistics and Probability, 12(1), 1–33.
  4. [4]Behme, A., Maejima, M., Matsui, M., & Sakuma, N. (2012). Distributions of exponential integrals of independent increment processes related to generalized gamma convolutions. Bernoulli, 18(4), 1172–1187. https://doi.org/10.3150/11-BEJ382
  5. [5]Bertoin, J., & Yor, M. (2005). Exponential functionals of Lévy processes. Probability surveys, 2, 191–212. https://doi.org/10.1214/154957805100000122
  6. [6]Bondesson, L. (1990). Generalized gamma convolutions and complete monotonicity. Probability Theory and Related Fields, 85, 181–194. https://doi.org/10.1007/BF01277981
  7. [7]Bondesson, L. (1992). Generalized gamma convolutions and related classes of distributions and densities (1st ed. 1992.). Springer New York.DOI
  8. [8]Bondesson, L. (2015). A class of probability distributions that is closed with respect to addition as well as multiplication of independent random variables. Journal of theoretical probability, 28(3), 1063–1081. https://doi.org/10.1007/s10959-013-0523-y
  9. [9]Bougerol, P. (1983). Exemples de théorèmes locaux sur les groupes résolubles. Annales de l’I.H.P. Probabilités et statistiques, 19(4), 369–391.
  10. [10]Choi, J., Liu, C., & Seo, B. K. (2019). Hyperbolic normal stochastic volatility model. The journal of futures markets, 39(2), 186–204. https://doi.org/10.1002/fut.21697
  11. [11]Dufresne, D. (1990). The distribution of a perpetuity, with applications to risk theory and pension funding. Scandinavian Actuarial Journal, 1990(1), 39–79. https://doi.org/10.1080/03461238.1990.10413872
  12. [12]Jedidi, W., & Vakeroudis, S. (2018). Windings of planar processes, exponential functionals and Asian options. Advances in applied probability, 50(3), 726–742. https://doi.org/10.1017/apr.2018.33
  13. [13]Johnson, N. L. (1949). Systems of frequency curves generated by methods of translation. Biometrika, 36(1/2), 149–176. https://doi.org/10.2307/2332539
  14. [14]Jurek, Z. J., & Vervaat, W. (1983). An integral representation for selfdecomposable Banach space valued random variables. Zeitschrift für Wahrscheinlichkeitstheorie und Verwandte Gebiete, 62(2), 247–262. https://doi.org/10.1007/BF00538800
  15. [15]Kondo, H., Maejima, M., & Sato, K.-i. (2006). Some properties of exponential integrals of Lévy processes and examples. Electronic Communications in Probability, 11, 291–303. https://doi.org/10.1214/ECP.v11-1232
  16. [16]Matsumoto, H., & Yor, M. (2005). Exponential functionals of Brownian motion, ii: Some related diffusion processes. Probability surveys, 2, 348–384. https://doi.org/10.1214/154957805100000168
  17. [17]Norberg, R. (1999). Ruin problems with assets and liabilities of diffusion type. Stochastic Processes and their Applications, 81(2), 255–269. https://doi.org/10.1016/S0304-4149(98)00103-3
  18. [18]Patra, S. (2021). Revisiting value-at-risk and expected shortfall in oil markets under structural breaks: The role of fat-tailed distributions. Energy economics, 101, 105452–. https://doi.org/10.1016/j.eneco.2021.105452
  19. [19]Paulsen, J. (1993). Risk theory in a stochastic economic environment. Stochastic Processes and their Applications, 46(2), 327–361. https://doi.org/10.1016/0304-4149(93)90010-2
  20. [20]Paulsen, J. (2008). Ruin models with investment income. Probability Surveys, 5, 416–434. https://doi.org/10.1214/08-PS134
  21. [21]Sato, K.-i. (2013). Lévy processes and infinitely divisible distributions (Revised edition.). Cambridge University Press.
  22. [22]Sato, K.-i., & Yamazato, M. (1983). Stationary processes of Ornstein-Uhlenbeck type. In J. V. Prokhorov & K. Itô (Eds.), Probability theory and mathematical statistics (pp. 541–551). Springer Berlin Heidelberg. https://doi.org/10.1007/BFb0072949
  23. [23]Steutel, F. W., & van Harn, K. (2004). Infinite divisibility of probability distributions on the real line. Marcel Dekker.DOI
  24. [24]Thorin, O. (1977a). On the infinite divisibility of the pareto distribution. Scandinavian Actuarial Journal, 1977, 31–40. https://doi.org/10.1080/03461238.1977.10405623
  25. [25]Thorin, O. (1977b). On the infinite divisibility of the lognormal distribution. Scandinavian Actuarial Journal, 1977, 121–148. https://doi.org/10.1080/03461238.1977.10405635
  26. [26]Thorin, O. (1978). An extension of the notion of a generalized Γ-convolution. Scandinavian Actuarial Journal, 1978(3), 141–149. https://doi.org/10.1080/03461238.1978.10432021
  27. [27]Torricelli, L. (2026). Thorin processes and their subordination. Scandinavian Actuarial Journal, 2026(1), 21–59. https://doi.org/10.1080/03461238.2025.2501062
  28. [28]Vakeroudis, S. (2012). Bougerol’s identity in law and extensions. Probability Surveys, 9, 411–437. https://doi.org/10.1214/12-PS195
  29. [29]Wolfe, S. J. (1982). On a continuous analogue of the stochastic difference equation Xₙ = ρXₙ₋₁+Bₙ. Stochastic processes and their applications, 12(3), 301–312. https://doi.org/10.1016/0304-4149(82)90050-3
  30. [30]Zhang, Y., Liu, H., Huang, Y., & Deng, Z. (2018). A nonlinear ARMA-GARCH model with Johnson Su innovations and its application to sea clutter modeling. IEEE access, 6, 11888–11896. https://doi.org/10.1109/ACCESS.2018.2805306

Paper details

vlnc:W_w24b69up
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents