Abstract
The multivariate skewed variance gamma (MSVG) distribution is useful in modelling data with high density around the location parameter along with moderate heavy-tailedness. However, the density can be unbounded for certain choices of shape parameter. We propose a modification to the expectation-conditional maximisation (ECM) algorithm to calculate the maximum likelihood estimate (MLE) by introducing a small region to cap the conditional expectations in order to deal with the unbounded density. To facilitate application to financial time series, the mean is further extended to include autoregressive terms. Finally, the MSVG model is applied to analyse the returns of five daily closing price market indices. Standard error (SE) for the estimated parameters are computed using Louis’ method.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
Aas K, Haff IH (2006) The generalized hyperbolic skew student’s t-distribution. J Financ Econom 4(2):275–309
Abramowitz M, Stegun IA (1964) Handbook of mathematical functions with formulas, graphs and mathematical tables. Dover, New York
Barndorff-Nielsen O (1978) Hyperbolic distributions and distributions on hyperbolae. Scand. J. Stat:151–157
Breymann W, Lüthi D (2013) ghyp: a package on generalized hyperbolic distributions. Available in http://cran.r-project.org/web/packages/ghyp/vignettes/Generalized_Hyperbolic_Distribution.pdf
Chou R (2005) Forecasting financial volatilities with extreme values: The conditional autoregressive range (carr) model. J Money Credit Banking 37(3):561–82
Choy STB, Chen CWS, Lin EMH (2014) Bivariate asymmetric garch models with heavy tails and dynamic conditional correlations. Quantitative Finance 14 (7):1297–1313
Dempster AP, Laird NM, Rubin DB (1977) Maximum likelihood from incomplete data via the EM algorithm. J Roy Statist Soc Ser B 39(1):1–38
Embrechts P (1983) A property of the generalized inverse Gaussian distribution with some applications. J Appl Probab 20(3):537–544
Finlay R, Seneta E (2008) Stationary-increment variance-gamma and “t” models: Simulation and parameter estimation. Int Statist Rev 76(2):167–186
Fung T, Seneta E (2007) Tailweight, quantiles and kurtosis: a study of competing distributions. Oper Res Lett 35(4):448–454
Hu W (2005) Calibration of multivariate generalized hyperbolic distributions using the EM algorithm, with applications in risk management, portfolio optimization and portfolio credit risk. Florida State University
Karlis D (2002) An em type algorithm for maximum likelihood estimation of the normal–inverse Gaussian distribution. Stat Prob Lett 57(1):43–52
Kawai R (2015) On the likelihood function of small time variance gamma Lévy processes. Statistics 49(1):63–83
Kotz S, Kozubowski TJ, Podgórski K (2001) The Laplace distribution and generalizations : a revisit with applications to communications, economics, engineering and finance. Birkhäuser, Boston
Liu C, Rubin DB (1994) The ECME algorithm: a simple extension of EM and ECM with faster monotone convergence. Biometrika 81(4):633–648
Liu C, Rubin DB (1995) ML estimation of the t distribution using EM and its extensions, ECM and ECME. Statist Sinica 5(1):19–39
Louis TA (1982) Finding the observed information matrix when using the EM algorithm. J Roy Statist Soc Ser B 44(2):226–233
Madan DB, Seneta E (1987) Chebyshev polynomial approximations and characteristic function estimation. J Roy Statist Soc Ser B 49(2):163–169
Madan DB, Seneta E (1990) The variance gamma (V.G.) model for share market returns. J Bus 63(4):511–524
Madan DB, Carr P, Chang EC (1998) The variance gamma process and option pricing. Eur Financ Rev 2:79–105
McNeil AJ, Frey R, Embrechts P (2005) Quantitative risk management. Princeton Series in Finance. Princeton University Press, Princeton, NJ. Concepts, techniques and tools
Meng X-L (1994) On the rate of convergence of the ECM algorithm. Ann Statist 22(1):326–339
Meng X-L, Rubin DB (1993) Maximum likelihood estimation via the ECM algorithm: a general framework. Biometrika 80(2):267–278
Meng X-L, Van Dyk D (1997) The em algorithm—an old folk-song sung to a fast new tune. J Roy Statist Soc Ser B 59(3):511–567
Nelder JA, Mead R (1965) A simplex method for function minimization. Comput J 7:308–313
Podgórski K, Wallin J (2015) Maximizing leave-one-out likelihood for the location parameter of unbounded densities. Ann Inst Statist Math 67(1):19–38
Protassov RS (2004) Em-based maximum likelihood parameter estimation for multivariate generalized hyperbolic distributions with fixed λ. Stat Comput 14(1):67–77
Seneta E (2004) Fitting the variance-gamma model to financial data. J Appl Probab 41A:177–187
Shao X-D, Lian Y-J, Yin L-Q (2009) Forecasting value-at-risk using high frequency data: the realized range model. Glob Financ J 20(2):128–136
Tjetjep A, Seneta E (2006) Skewed normal variance-mean models for asset pricing and the method of moments. Int Statist Rev 74(1):109–126
Author information
Authors and Affiliations
Corresponding author
Additional information
Complementary Materials
See https://github.sydney.edu.au/tnit9903/ECM-Algo-for-AR-MSVG for the Appendix and the code for the simulations and implementations in Sections 4 and 5.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Nitithumbundit, T., Chan, J.S.K. ECM Algorithm for Auto-Regressive Multivariate Skewed Variance Gamma Model with Unbounded Density. Methodol Comput Appl Probab 22, 1169–1191 (2020). https://doi.org/10.1007/s11009-019-09762-0
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11009-019-09762-0
Keywords
- Unbounded likelihood
- Multivariate skewed variance gamma distribution
- EM algorithm
- Observed information matrix
- Normal mean-variance representation