Abstract
We perform an extensive investigation of different specifications of the BEKK-type multivariate volatility models for a moderate number of assets, focusing on how the degree of parametrization affects forecasting performance. Because the unrestricted specification may be too generously parameterized, often one imposes restrictions on coefficient matrices constraining them to have a diagonal or even scalar structure. We frame all three model variations (full, diagonal, scalar) as special cases of a ridge-type regularized estimator, where the off-diagonal elements are shrunk towards zero and the diagonal elements are shrunk towards homogeneity. Our forecasting experiments with BEKK-type Conditional Autoregressive Wishart model for realized volatility confirm the superiority of the more parsimonious scalar and diagonal model variations. Even though sometimes a moderate degree of regularization of the diagonal and off-diagonal parameters may be beneficial for forecasting performance, it does not regularly lead to tangible performance improvements irrespective of how precise is tuning of regularization intensity. Additionally, our results highlight the crucial importance of frequent model re-estimation in improving the forecast precision, and, perhaps paradoxically, a slight advantage of shorter estimation windows compared to longer windows.
Funding source: Grantová Agentura České Republiky http://dx.doi.org/10.13039/501100001824
Award Identifier / Grant number: 20-28055S
Acknowledgment
We thank an anonymous referee for the helpful comments and constructive remarks. Furthermore, we thank the audiences and discussants at the 40th International Symposium on Forecasting, XIt Workshop in Time Series Econometrics, and 5th International Workshop on Financial Markets and Nonlinear Dynamics.
-
Author contribution: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.
-
Research funding: Czech Science Foundation support under grant 20-28055S and Grantová agentura UK support under grant 264120 are gratefully acknowledged.
-
Conflict of interest statement: The authors declare no conflicts of interest regarding this article.
Figures 5–7 and Tables 5 and 6
n = 2 | n = 3 | n = 4 | n = 5 | n = 6 | n = 7 | n = 8 | n = 9 | n = 10 | All | |
---|---|---|---|---|---|---|---|---|---|---|
(A) sCAW | ||||||||||
2y | 2.11 | 2.24 | 2.23 | 2.17 | 1.96 | 1.97 | 2.00 | 1.94 | 1.89 | 2.06 |
3y | 2.21 | 2.38 | 2.46 | 2.26 | 2.10 | 2.18 | 2.11 | 1.99 | 2.14 | 2.20 |
4y | 2.61 | 2.61 | 2.88 | 2.60 | 2.40 | 2.54 | 2.47 | 2.14 | 1.78 | 2.49 |
5y | 2.88 | 2.77 | 3.07 | 3.18 | 3.18 | 3.12 | 3.00 | 2.94 | 3.07 | 3.01 |
All | 2.42 | 2.47 | 2.60 | 2.49 | 2.35 | 2.39 | 2.35 | 2.19 | 2.16 | 2.39 |
(B) dCAW | ||||||||||
2y | 2.47 | 2.72 | 2.84 | 2.78 | 2.88 | 2.87 | 2.96 | 2.92 | 3.00 | 2.82 |
3y | 2.36 | 2.50 | 2.80 | 2.89 | 2.80 | 2.96 | 3.04 | 3.08 | 3.09 | 2.83 |
4y | 2.44 | 2.53 | 2.86 | 2.56 | 2.81 | 2.94 | 2.94 | 3.11 | 3.22 | 2.81 |
5y | 2.39 | 2.32 | 2.54 | 2.57 | 2.57 | 2.71 | 2.50 | 2.68 | 2.71 | 2.55 |
All | 2.43 | 2.59 | 2.78 | 2.72 | 2.81 | 2.84 | 2.86 | 2.97 | 3.04 | 2.77 |
(C) fCAW | ||||||||||
2y | 2.98 | 2.84 | 2.76 | 2.84 | 3.04 | 3.17 | 3.12 | 3.36 | 3.39 | 3.05 |
3y | 2.81 | 2.93 | 2.48 | 2.77 | 3.02 | 2.93 | 2.91 | 3.06 | 3.00 | 2.88 |
4y | 2.36 | 2.47 | 2.08 | 2.53 | 2.67 | 2.50 | 2.44 | 2.48 | 2.89 | 2.46 |
5y | 2.21 | 2.36 | 2.11 | 2.00 | 1.82 | 2.07 | 2.21 | 2.34 | 2.29 | 2.16 |
All | 2.63 | 2.64 | 2.41 | 2.59 | 2.69 | 2.74 | 2.76 | 2.83 | 2.91 | 2.68 |
(D) rCAW | ||||||||||
2y | 2.45 | 2.20 | 2.17 | 2.21 | 2.12 | 1.99 | 1.92 | 1.78 | 1.73 | 2.07 |
3y | 2.62 | 2.19 | 2.27 | 2.08 | 2.08 | 1.93 | 1.93 | 1.88 | 1.77 | 2.10 |
4y | 2.58 | 2.39 | 2.18 | 2.32 | 2.12 | 2.01 | 2.14 | 2.27 | 2.11 | 2.25 |
5y | 2.52 | 2.55 | 2.29 | 2.25 | 2.43 | 2.09 | 2.29 | 2.04 | 1.93 | 2.28 |
All | 2.52 | 2.30 | 2.21 | 2.20 | 2.15 | 2.04 | 2.03 | 2.00 | 1.89 | 2.16 |
-
Bold values refer to average rankings across all asset numerosities or all window lengths.
n = 2 | n = 3 | n = 4 | n = 5 | n = 6 | n = 7 | n = 8 | n = 9 | n = 10 | All | |
---|---|---|---|---|---|---|---|---|---|---|
(A) sCAW | ||||||||||
2y | 0.11 | 0.12 | 0.14 | 0.14 | 0.02 | 0.14 | 0.15 | 0.12 | 0.08 | 0.12 |
3y | 0.02 | 0.20 | 0.14 | 0.14 | 0.09 | 0.07 | 0.14 | 0.06 | 0.09 | 0.10 |
4y | 0.11 | 0.17 | 0.14 | 0.06 | 0.03 | 0.06 | 0.03 | 0.04 | 0.00 | 0.07 |
5y | 0.07 | 0.07 | 0.11 | 0.18 | 0.25 | 0.21 | 0.18 | 0.22 | 0.29 | 0.17 |
All | 0.07 | 0.14 | 0.15 | 0.12 | 0.10 | 0.11 | 0.13 | 0.10 | 0.10 | 0.11 |
(B) dCAW | ||||||||||
2y | 0.21 | 0.30 | 0.26 | 0.18 | 0.22 | 0.36 | 0.33 | 0.33 | 0.46 | 0.28 |
3y | 0.07 | 0.20 | 0.14 | 0.16 | 0.16 | 0.14 | 0.16 | 0.20 | 0.09 | 0.15 |
4y | 0.17 | 0.11 | 0.08 | 0.06 | 0.03 | 0.06 | 0.08 | 0.07 | 0.11 | 0.08 |
5y | 0.11 | 0.04 | 0.07 | 0.11 | 0.14 | 0.18 | 0.04 | 0.10 | 0.14 | 0.10 |
All | 0.14 | 0.19 | 0.16 | 0.13 | 0.15 | 0.19 | 0.17 | 0.19 | 0.20 | 0.17 |
(C) fCAW | ||||||||||
2y | 0.23 | 0.24 | 0.28 | 0.26 | 0.37 | 0.31 | 0.29 | 0.36 | 0.38 | 0.30 |
3y | 0.17 | 0.23 | 0.11 | 0.20 | 0.23 | 0.30 | 0.23 | 0.30 | 0.18 | 0.22 |
4y | 0.17 | 0.17 | 0.06 | 0.08 | 0.17 | 0.06 | 0.08 | 0.09 | 0.11 | 0.11 |
5y | 0.11 | 0.14 | 0.07 | 0.04 | 0.07 | 0.11 | 0.11 | 0.15 | 0.00 | 0.10 |
All | 0.16 | 0.20 | 0.13 | 0.16 | 0.22 | 0.19 | 0.19 | 0.23 | 0.23 | 0.19 |
(D) rCAW | ||||||||||
2y | 0.17 | 0.14 | 0.18 | 0.20 | 0.20 | 0.19 | 0.19 | 0.18 | 0.23 | 0.18 |
3y | 0.10 | 0.16 | 0.14 | 0.11 | 0.09 | 0.07 | 0.07 | 0.08 | 0.00 | 0.10 |
4y | 0.14 | 0.11 | 0.06 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.04 |
5y | 0.14 | 0.14 | 0.11 | 0.11 | 0.11 | 0.14 | 0.07 | 0.15 | 0.14 | 0.12 |
All | 0.14 | 0.15 | 0.13 | 0.11 | 0.10 | 0.11 | 0.10 | 0.10 | 0.10 | 0.12 |
-
Bold values refer to average rankings across all asset numerosities or all window lengths.
References
Alfelt, G., T. Bodnar, and J. Tyrcha. 2020. “Goodness-of-Fit Tests for Centralized Wishart Processes.” Communications in Statistics - Theory and Methods 49 (20): 5060–90. https://doi.org/10.1080/03610926.2019.1612917.Search in Google Scholar
Anatolyev, S. 2020. “A Ridge to Homogeneity for Linear Models.” Journal of Statistical Computation and Simulation 90 (13): 2455–72. https://doi.org/10.1080/00949655.2020.1779722.Search in Google Scholar
Anatolyev, S., and N. Kobotaev. 2018. “Modeling and Forecasting Realized Covariance Matrices with Accounting for Leverage.” Econometric Reviews 37 (2): 114–39. https://doi.org/10.1080/07474938.2015.1035165.Search in Google Scholar
Andersen, T. G., T. Bollerslev, F. X. Diebold, and P. Labys. 2003. “Modeling and Forecasting Realized Volatility.” Econometrica 71 (2): 579–625. https://doi.org/10.1111/1468-0262.00418.Search in Google Scholar
Anderson, T. W., and D. A. Darling. 1952. “Asymptotic Theory of Certain “Goodness of Fit” Criteria Based on Stochastic Processes.” The Annals of Mathematical Statistics 23 (2): 193–212. https://doi.org/10.1214/aoms/1177729437.Search in Google Scholar
Asai, M., R. Gupta, and M. McAleer. 2020. “Forecasting Volatility and Co-Volatility of Crude Oil and Gold Futures: Effects of Leverage, Jumps, Spillovers, and Geopolitical Risks.” International Journal of Forecasting 36 (3): 933–48. https://doi.org/10.1016/j.ijforecast.2019.10.003.Search in Google Scholar
Barndorff-Nielsen, O. E., and N. Shephard. 2004. “Econometric Analysis of Realized Covariation: High Frequency Based Covariance, Regression, and Correlation in Financial Economics.” Econometrica 72 (3): 885–925. https://doi.org/10.1111/j.1468-0262.2004.00515.x.Search in Google Scholar
Bollerslev, T. 1990. “Modelling the Coherence in Short-Run Nominal Exchange Rates: A Multivariate Generalized ARCH Model.” The Review of Economics and Statistics 72 (3): 498–505. https://doi.org/10.2307/2109358.Search in Google Scholar
Box, G. E. P., and D. A. Pierce. 1970. “Distribution of Residual Autocorrelations in Autoregressive-Integrated Moving Average Time Series Models.” Journal of the American Statistical Association 65 (332): 1509–26. https://doi.org/10.1080/01621459.1970.10481180.Search in Google Scholar
Caporin, M., and M. McAleer. 2012. “Do We Really Need Both BEKK and DCC? A Tale of Two Multivariate GARCH Models.” Journal of Economic Surveys 26 (4): 736–51. https://doi.org/10.1111/j.1467-6419.2011.00683.x.Search in Google Scholar
Caporin, M., and M. McAleer. 2014. “Robust Ranking of Multivariate GARCH Models by Problem Dimension.” Computational Statistics & Data Analysis 76: 172–85. https://doi.org/10.1016/j.csda.2012.05.012.Search in Google Scholar
Caporin, M., and P. Paruolo. 2015. “Proximity-Structured Multivariate Volatility Models.” Econometric Reviews 34 (5): 559–93. https://doi.org/10.1080/07474938.2013.807102.Search in Google Scholar
Clark, T., and M. McCracken. 2013. “Advances in Forecast Evaluation.” Handbook of Economic Forecasting 2: 1107–201. https://doi.org/10.1016/b978-0-444-62731-5.00020-8.Search in Google Scholar
Engle, R. F. 2002. “Dynamic Conditional Correlation: A Simple Class of Multivariate Generalized Autoregressive Conditional Heteroskedasticity Models.” Journal of Business & Economic Statistics 20 (3): 339–50. https://doi.org/10.1198/073500102288618487.Search in Google Scholar
Engle, R. F., and K. F. Kroner. 1995. “Multivariate Simultaneous Generalized ARCH.” Econometric Theory 11 (1): 122–50. https://doi.org/10.1017/s0266466600009063.Search in Google Scholar
Golosnoy, V., B. Gribisch, and R. Liesenfeld. 2012. “The Conditional Autoregressive Wishart Model for Multivariate Stock Market Volatility.” Journal of Econometrics 167 (1): 211–23. https://doi.org/10.1016/j.jeconom.2011.11.004.Search in Google Scholar
Hansen, P. R., A. Lunde, and J. M. Nason. 2011. “The Model Confidence Set.” Econometrica 79 (2): 453–97.10.3982/ECTA5771Search in Google Scholar
Hoerl, A. E., and R. W. Kennard. 1970. “Ridge Regression: Biased Estimation for Nonorthogonal Problems.” Technometrics 12 (1): 55–67. https://doi.org/10.1080/00401706.1970.10488634.Search in Google Scholar
James, W., and C. Stein. 1961. “Estimation with Quadratic Loss.” In Proceedings of the Fourth Berkeley Symposium on Mathematical Statistics and Probability 1961, 443–60.Search in Google Scholar
Laurent, S., J. V. Rombouts, and F. Violante. 2012. “On the Forecasting Accuracy of Multivariate GARCH Models.” Journal of Applied Econometrics 27 (6): 934–55. https://doi.org/10.1002/jae.1248.Search in Google Scholar
Lucheroni, C., J. Boland, and C. Ragno. 2019. “Scenario Generation and Probabilistic Forecasting Analysis of Spatio-Temporal Wind Speed Series with Multivariate Autoregressive Volatility Models.” Applied Energy 239: 1226–41. https://doi.org/10.1016/j.apenergy.2019.02.015.Search in Google Scholar
Noureldin, D., N. Shephard, and K. Sheppard. 2012. “Multivariate High-Frequency-Based Volatility (HEAVY) Models.” Journal of Applied Econometrics 27 (6): 907–33. https://doi.org/10.1002/jae.1260.Search in Google Scholar
Patton, A. J. 2011. “Volatility Forecast Comparison Using Imperfect Volatility Proxies.” Journal of Econometrics 160: 246–56. https://doi.org/10.1016/j.jeconom.2010.03.034.Search in Google Scholar
Sheppard, K. 2013. MFE Toolbox. Also available at https://www.kevinsheppard.com/code/matlab/mfe-toolbox.Search in Google Scholar
Zhipeng, Y., and L. Shenghong. 2018. “Hedge Ratio on Markov Regime-Switching Diagonal BEKK–GARCH Model.” Finance Research Letters 24: 49–55. https://doi.org/10.1016/j.frl.2017.06.015.Search in Google Scholar
Zolfaghari, M., H. Ghoddusi, and F. Faghihian. 2020. “Volatility Spillovers for Energy Prices: A Diagonal BEKK Approach.” Energy Economics 92: 104965. https://doi.org/10.1016/j.eneco.2020.104965.Search in Google Scholar
Supplementary Material
The online version of this article offers supplementary material (https://doi.org/10.1515/snde-2021-0064).
© 2022 Walter de Gruyter GmbH, Berlin/Boston