Hostname: page-component-8448b6f56d-jr42d Total loading time: 0 Render date: 2024-04-20T03:23:31.545Z Has data issue: false hasContentIssue false
  • English
  • Français

Marshall–Olkin distributions, subordinators, efficient simulation, and applications to credit risk

Part of: Markov processes Stochastic processes

Published online by Cambridge University Press:  26 June 2017

Yunpeng Sun ,
Rafael Mendoza-Arriaga  and
Vadim Linetsky
Yunpeng Sun*
Affiliation:
Northwestern University
Rafael Mendoza-Arriaga*
Affiliation:
The University of Texas at Austin
Vadim Linetsky*
Affiliation:
Northwestern University
*
* Postal address: Department of Industrial Engineering and Management Sciences, McCormick School of Engineering and Applied Sciences, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA.
*** Postal address: Department of Information, Risk, and Operations Management, McCombs School of Business, The University of Texas at Austin, CBA 5.202, B6500, 1 University Station, Austin, TX 78712, USA. Email address: rafael.mendoza-arriaga@mccombs.utexas.edu
* Postal address: Department of Industrial Engineering and Management Sciences, McCormick School of Engineering and Applied Sciences, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA.
Get access Rights & Permissions [Opens in a new window]

Abstract

In the paper we present a novel construction of Marshall–Olkin (MO) multivariate exponential distributions of failure times as distributions of the first-passage times of the coordinates of multidimensional Lévy subordinator processes above independent unit-mean exponential random variables. A time-inhomogeneous version is also given that replaces Lévy subordinators with additive subordinators. An attractive feature of MO distributions for applications, such as to portfolio credit risk, is its singular component that yields positive probabilities of simultaneous defaults of multiple obligors, capturing the default clustering phenomenon. The drawback of the original MO fatal shock construction of MO distributions is that it requires one to simulate 2n-1 independent exponential random variables. In practice, the dimensionality is typically on the order of hundreds or thousands of obligors in a large credit portfolio, rendering the MO fatal shock construction infeasible to simulate. The subordinator construction reduces the problem of simulating a rich subclass of MO distributions to simulating an n-dimensional subordinator. When one works with the class of subordinators constructed from independent one-dimensional subordinators with known transition distributions, such as gamma and inverse Gaussian, or their Sato versions in the additive case, the simulation effort is linear in n. To illustrate, we present a simulation of 100,000 samples of a credit portfolio with 1,000 obligors that takes less than 18 seconds on a PC.

Keywords

Marshall–Olkin multivariate exponential distributionLévy processadditive subordinatorsubordinatorsimulationdependent lifetimefailuredefaultreliabilitycredit risk

MSC classification

Primary: 60G51: Processes with independent increments; L'evy processes
Secondary: 60J28: Applications of continuous-time Markov processes on discrete state spaces
Type
Research Article
Information
Advances in Applied Probability , Volume 49 , Issue 2 , June 2017 , pp. 481 - 514
Copyright
Copyright © Applied Probability Trust 2017 

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

[1] Aven, T. and Jensen, U. (1999). Stochastic Models in Reliability (Appl. Math. 41). Springer, New York. CrossRefGoogle Scholar
[2] Barlow, R. E. and Proschan, F. (1975). Statistical Theory of Reliability and Life Testing: Probability Models. Holt, Rinehart and Winston, New York. Google Scholar
[3] Barndorff-Nielsen, O. E. (1998). Processes of normal inverse Gaussian type. Finance Stoch. 2, 41–68. CrossRefGoogle Scholar
[4] Barndorff-Nielsen, O. E., Pedersen, J. and Sato, K. (2001). Multivariate Subordination, Self-Decomposability and Stability. Adv. Appl. Prob. 33, 160–187. CrossRefGoogle Scholar
[5] Bertoin, J. (1996). Lévy Processes (Camb. Tracts Math. 121). Cambridge University Press. Google Scholar
[6] Bielecki, T. R. and Rutkowski, M. (2004). Credit Risk: Modeling, Valuation and Hedging. Springer, Berlin. CrossRefGoogle Scholar
[7] Bielecki, T. R., Cousin, A., Crépey, S. and Herbertsson, A. (2014). Dynamic hedging of portfolio credit risk in a Markov copula model. J. Optimization Theory Appl. 161, 90–102. CrossRefGoogle Scholar
[8] Carr, P., Geman, H., Madan, D. B. and Yor, M. (2003). Stochastic volatility for Lévy processes. Math. Finance 13, 345–382. CrossRefGoogle Scholar
[9] Çinlar, E. and Özekici, S. (1987). Reliability of complex devices in random environments. Prob. Eng. Inf. Sci. 1, 97–115. CrossRefGoogle Scholar
[10] Çinlar, E., Shaked, M. and Shanthikumar, J. G. (1989). On lifetimes influenced by a common environment. Stoch. Process. Appl. 33, 347–359. CrossRefGoogle Scholar
[11] Cheng, R. C. H. (1977). The generation of gamma variables with non-integral shape parameter. J. R. Statist. Soc. C 26, 71–75. Google Scholar
[12] Cont, R. and Tankov, P. (2004). Financial Modelling with Jump Processes. Chapmann & Hall/CRC, Boca Raton, FL. Google Scholar
[13] Crépey, S., Jeanblanc, M. and Zargari, B. (2010). Counterparty risk on a CDS in a Markov chain copula model with joint defaults. In Recent Advances in Financial Engineering, eds M. Kijima et al., World Scientific, Singapore, pp. 91–126. Google Scholar
[14] Duffie, D. and Gârleanu, N. (2001). Risk and valuation of collateralized debt obligations. Financial Anal. J. 57, 41–59. CrossRefGoogle Scholar
[15] Eberlein, E. and Madan, D. B. (2009). Sato processes and the valuation of structured products. Quant. Finance 9, 27–42. CrossRefGoogle Scholar
[16] Elouerkhaoui, Y. (2007). Pricing and hedging in a dynamic credit model. Internat. J. Theoret. Appl. Finance 10, 703–731. CrossRefGoogle Scholar
[17] Frees, E. W., Carriere, J. and Valdez, E. (1996). Annuity valuation with dependent mortality. J. Risk Insurance 63, 229–261. CrossRefGoogle Scholar
[18] Gaspar, R. M. and Schmidt, T. (2010). Credit risk modelling with shot-noise processes. Working paper. CrossRefGoogle Scholar
[19] Ghosh, S. K. and Gelfand, A. E. (1998). Latent risk approach for modeling individual level data consisting of multiple event times. J. Statist. Res. 32, 23–38. Google Scholar
[20] Giesecke, K. (2003). A simple exponential model for dependent defaults. J. Fixed Income 13, 74–83. CrossRefGoogle Scholar
[21] Glasserman, P. (2004). Monte Carlo Methods in Financial Engineering (Appl. Math. 53). Springer, New York. Google Scholar
[22] Gradshteyn, I. S. and Ryzhik, I. M. (2000). Table of Integrals, Series, and Products, 6th edn. Academic Press, San Diego, CA. Google Scholar
[23] Harris, R. (1968). Reliability applications of a bivariate exponential distribution. Operat. Res. 16, 18–27. CrossRefGoogle Scholar
[24] Herbertsson, A., Jang, J. and Schmidt, T. (2011). Pricing basket default swaps in a tractable shot noise model. Statist. Prob. Lett. 81, 1196–1207. CrossRefGoogle Scholar
[25] Hofert, M. and Vrins, F. (2013). Sibuya copulas. J. Multivariate Anal. 114, 318–337. CrossRefGoogle Scholar
[26] Jeanblanc, M., Yor, M. and Chesney, M. (2009). Mathematical Methods for Financial Markets. Springer, London. CrossRefGoogle Scholar
[27] Joshi, M. S. and Stacey, A. M. (2006). Intensity gamma: a new approach to pricing portfolio credit derivatives. Risk Mag. 78–83. Google Scholar
[28] Klein, J. P., Keiding, N. and Kamby, C. (1989). Semiparametric Marshall–Olkin models applied to the occurrence of metastases at multiple sites after breast cancer. Biometrics 45, 1073–1086. CrossRefGoogle Scholar
[29] Kokholm, T. and Nicolato, E. (2010). Sato processes in default modelling. Appl. Math. Finance 17, 377–397. CrossRefGoogle Scholar
[30] Kou, S. and Peng, X. H. (2008). Connecting the top-down to the bottom-up: pricing CDO under a conditional survival (CS) model. In Proc. 2008 Winter Simulation Conference, IEEE, pp. 578–586. Google Scholar
[31] Li, L. and Mendoza-Arriaga, R. (2013). Ornstein-Uhlenbeck processes time changed with additive subordinators and their applications in commodity derivative models. Operat. Res. Lett. 41, 521–525. CrossRefGoogle Scholar
[32] Li, J., Li, L. and Mendoza-Arriaga, R. (2016). Additive subordination and its applications in finance. Finance Stoch. 20, 589–634. CrossRefGoogle Scholar
[33] Lindskog, F. and McNeil, A. J. (2003). Common Poisson shock models: applications to insurance and credit risk modelling. ASTIN Bull. 33, 209–238. CrossRefGoogle Scholar
[34] Madan, D. B. and Seneta, E. (1990). The variance gamma (V.G.) model for share market returns. J. Business 63, 511–524. CrossRefGoogle Scholar
[35] Madan, D. B., Carr, P. P. and Chang, E. C. (1998). The variance gamma process and option pricing. Rev. Finance 2, 79–105. CrossRefGoogle Scholar
[36] Mai, J.-F. and Scherer, M. (2009). Lévy-frailty copulas. J. Multivariate Anal. 100, 1567–1585. CrossRefGoogle Scholar
[37] Mai, J.-F. and Scherer, M. (2009). A tractable multivariate default model based on a stochastic time-change. Internat. J. Theoret. Appl. Finance 12, 227–249. CrossRefGoogle Scholar
[38] Mai, J.-F. and Scherer, M. (2011). Reparameterizing Marshall–Olkin copulas with applications to sampling. J. Statis. Comput. Simul. 81, 59–78. CrossRefGoogle Scholar
[39] Mai, J.-F. and Scherer, M. A. (2013). Extendibility of Marshall–Olkin distributions and inverse Pascal triangles. Brazilian J. Prob. Statist. 27, 310–321. CrossRefGoogle Scholar
[40] Manistre, B. J. and Hancock, G. H. (2005). Variance of the CTE estimator. N. Amer. Actuarial J. 9, 129–156. CrossRefGoogle Scholar
[41] Marsaglia, G. and Tsang, W. W. (2000). The Ziggurat method for generating random variables. J. Statist. Software 5, 1–7. CrossRefGoogle Scholar
[42] Marshall, A. W. and Olkin, I. (1967). A multivariate exponential distribution. J. Amer. Statist. Assoc. 62, 30–44. CrossRefGoogle Scholar
[43] Mendoza-Arriaga, R. and Linetsky, V. (2016). Multivariate subordination of Markov processes with financial applications. Math. Finance 26, 699–747. CrossRefGoogle Scholar
[44] Mendoza-Arriaga, R., Carr, P. and Linetsky, V. (2010). Time-changed Markov processes in credit-equity modeling. Math. Finance 20, 527–569. CrossRefGoogle Scholar
[45] Sato, K.-I. (1999). Lévy Processes and Infinitely Divisible Distributions (Camb. Stud. Adv. Math. 68). Cambridge University Press. Google Scholar
[46] Scherer, M., Schmid, L. and Schmidt, T. (2012). Shot-noise driven multivariate default models. Europ. Actuarial J. 2, 161–186. CrossRefGoogle Scholar
[47] Schilling, R. L., Song, R. and Vondracek, Z. (2010). Bernstein Functions: Theory and Applications (De Gruyter Stud. Math. 37). De Gruyter, Berlin. Google Scholar
[48] Shaked, M. and Shanthikumar, J. G. (1986). Multivariate imperfect repair. Operat. Res. 34, 437–448. CrossRefGoogle Scholar
[49] Sun, Y., Mendoza-Arriaga, R. and Linetsky, V. (2016). Online appendix: Marshall–Olkin distributions, subordinators, efficient simulation, and applications to credit risk. Supplementary material. Available at http://doi.org/10.1017/apr.2017.10. CrossRefGoogle Scholar
[50] Vesely, W. (1977). Estimating common cause failure probabilities in reliability and risk analyses: Marshall–Olkin specializations. In Nuclear Systems Reliability Engineering and Risk Assessment: Papers, eds J. B. Fussell and G. R. Burdick, Society for Industrial and Applied Mathematics, Philadelphia, PA, pp. 314–341. Google Scholar
[51] Weiss, S. (2009). The inclusion exclusion principle and its more general version. Available at http://www.compsci.hunter.cuny.edu/~sweiss/resources/inclusion_exclusion.pdf. Google Scholar
Supplementary material: PDF

Sun supplementary material

Online Appendix

Download Sun supplementary material(PDF)
PDF 259.3 KB