Abstract
In this paper we consider mixture generalized autoregressive conditional heteroskedastic models, and propose a new iteration algorithm of type EM for the estimation of model parameters. The maximum likelihood estimates are shown to be consistent, and their asymptotic properties are investigated. More precisely, we derive simple expressions in closed form for the asymptotic covariance matrix and the expected Fisher information matrix of the ML estimator. Finally, we study the model selection and propose testing procedures. A simulation study and an application to financial real-series illustrate the results.
Similar content being viewed by others
References
Abramson A, Cohen I (2007) On the stationarity of Markov-switching GARCH processes. Econom Theory 23(3):485–500
Alexander C, Lazar E (2006) Normal mixture GARCH\((1, 1)\): applications to exchange rate modelling. J Appl Econom 21(3):307–336
Augustyniak M, Boudreault M, Morales M (2018) Maximum likelihood estimation of the Markov-switching GARCH model based on a general collapsing procedure. Methodol Comput Appl Probab 20(1):165–188
Bauwens L, Hafner CM, Rombouts JVK (2007) Multivariate mixed normal conditional heteroskedasticity. Comput Stat Data Anal 51:3551–3566
Bauwens L, Preminger A, Rombouts JVK (2010) Theory and inference for a Markov switching GARCH model. Econom J 13:218–244
Bibi A (2019) QML Estimation of asymmetric Markov switching GARCH\((p, q)\) processes. Commun Math Stat. https://doi.org/10.1007/s40304-019-00197-0
Bibi A, Ghezal A (2018) Markov switching bilinear GARCH models: structure and estimation. Commun Stat Theory Methods 47(2):307–323
Bibi A, Ghezal A (2019) QMLE of periodic time-varying bilinear GARCH models. Commun Stat Theory Methods 48(13):3291–3310
Bickel P, Ritov Y, Ryden T (1998) Asymptotic normality of the maximum likelihood estimator for general hidden Markov models. Ann Stat 26:1614–1635
Billio M, Cavicchioli M (2017) Markov switching GARCH models: filtering, approximations and duality. In: Corazza M, Legros F, Perna C, Sibillo M (eds) Math Stat Methods Actuar Sci Finance. Springer, Cham
Bollerslev T (1986) Generalized autoregressive conditional heteroskedasticity. J Econom 31:307–327
Cavicchioli M (2014a) Determining the number of regimes in Markov switching VAR and VMA models. J Times Ser Anal 35(2):173–186
Cavicchioli M (2014b) Autocovariance and linear transformations of Markov switching VARMA processes. Central Eur J Econ Model Econom 6:275–289
Cavicchioli M (2018) On mixture autoregressive conditional heteroskedasticity. J Stat Plan Inference 197:35–50
Chen CWS, Lee S, Chen SY (2016) Local non-stationarity test in mean for Markov switching GARCH models: an approximate Bayesian approach. Comput Stat 31:1–24
Cheng X, Yu PLH, Li WK (2009) On a dynamic mixture GARCH model. J Forecast 28:247–265
Choi BS (1992) ARMA model identification. Springer, New York
Chuffart T (2015) Selection criteria in regime switching conditional volatility models. Econometrics 3:289–316
Das D, Yoo BH (2004) A Bayesian MCMC algorithm for Markov switching GARCH models. In: Series Econometric Society, No. 179, North American Summer Meetings
Douc R, Moulines E, Ryden T (2004) Asymptotic properties of the maximum likelihood estimator in autoregressive models with Markov regime. Ann Stat 32:2254–2304
Engle RF (1982) Autoregressive conditional heteroskedasticity with estimates of the variance of the United Kingdom inflation. Econometrica 50:987–1007
Francq C, Roussignol M, Zakoïan JM (2001) Conditional heteroskedasticity driven by hidden Markov chains. J Time Ser Anal 22(2):197–220
Francq C, Zakoïan JM (2005) The \(L^2\)-structures of standard and switching-regime GARCH models. Stoch Process Appl 115:1557–1582
Francq C, Zakoïan JM (2008) Deriving the autocovariances of powers of Markov-switching GARCH models, with applications to statistical inference. Comput Stat Data Anal 52:3027–3046
Francq C, Zakoïan JM (2012) Strict stationarity and stationary generalized autoregressive conditional heteroscedasticity models. Econometrica 80(2):821–861
Frühwirth-Schnatter S, Celeux G, Robert CP, eds. (2019) Handbook of Mixture Analysis. CRC Press, Taylor & Francis Group, Boca-Raton, London-New York
Gable J, Van Norden S, Vigfusson R (1997) Analytical derivatives for Markov switching models. Comput Econ 10:187–194
Gerlach R, Tuyl F (2006) MCMC methods for comparing stochastic volatility and GARCH models. Int J Forecast 22:91–107
Glodek M, Schels M, Schwenker F (2013) Ensemble Gaussian mixture models for probability density estimation. Comput Stat 28:127–138
Haas M (2010) Skew-normal mixture and Markov-switching GARCH processes. Stud Nonlinear Dyn Econom 14(4):1–56
Haas M, Mittnik S, Paolella MS (2004a) A new approach to Markov-switching GARCH models. J Financ Econom 2(4):493–530
Haas M, Mittnik S, Paolella MS (2004b) Mixed normal conditional heteroskedasticity. J Financ Econom 2(2):211–250
Hamadeh T, Zakoïan JM (2011) Asymptotic properties of LS and QML estimators for a class of nonlinear GARCH processes. J Stat Plan Inference 141:488–507
Hamilton JD (1994) Time series analysis. Princeton University Press, Princeton
Hennecke JS, Rachev ST, Fabozzi FJ, Nikolov M (2011) MCMC based estimation of Markov switching ARMA–GARCH models. Appl Econ 43(3):259–271
Krolzig HM (1997) Markov-switching vector autoregressions: modelling, statistical inference and application to business cycle analysis. Springer, Berlin
Lee O (2013) The functional central limit theorem for ARMA–GARCH processes. Econom Lett 121:432–435
Liu JC (2006) Stationarity of a Markov-switching GARCH model. J Financ Econom 4(4):573–593
Lütkepohl H (2007) New introduction to multiple time series analysis, 2nd edn. Springer, Berlin
Magnus JR, Neudecker H (1986) Symmetry, 0–1 matrices and Jacobians. A review. Econom Theory 2:157–190
McLachlan G, Krishnan T (1997) The EM algorithm and extensions. Wiley, New York
McLachlan G, Peel D (2000) Finite mixture models. Wiley, New York
Tse YK, Tsui AKC (2002) A multivariate generalized autoregressive conditional heteroscedasticity model with time-varying correlations. J Bus Econ Stat 20(3):351–362
Zhang Z, Li WK, Yuen KC (2006) On a mixture GARCH time-series model. J Time Ser Anal 27(4):577–597
Wong CS, Li WK (2000) On a mixture autoregressive model. J R Stat Soc B 62:95–115
Wong CS, Li WK (2001) On a mixture autoregressive conditional heteroskedastic model. J Am Stat Assoc 96:982–995
Acknowledgements
This work is financially supported by a FAR 2020 research grant of the University of Modena and Reggio E., Italy. We should like to thank the Editor in Chief of the journal, Professor Wataru Sakamoto, and two anonymous referees for their constructive comments and very useful suggestions and remarks which were most valuable for improvement of the final version of the paper.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix
Appendix
Proof of Proposition 3.1
Let \(f(y_t | \mathbf{z}_t, \mathbf{Y}_{t - 1}; {\varvec{\lambda }})\) denote the conditional density, and \(Pr(\mathbf{z}_t | \mathbf{z}_{t - 1}; {\varvec{\lambda }})\) the conditional probability, for every \(t = 1, \dots , T\).
Set
and
Then the likelihood function can be written as
where the integration denotes summation over all possible values of \(\mathbf{z} \, = \, \mathbf{z}_T \otimes \mathbf{z}_{T - 1} \otimes \cdots \otimes \mathbf{z}_1\). Here \(\otimes \) is the usual Kronecker product. The derivation of the log likelihood function \(\mathcal{L}({\varvec{\lambda }} | \mathbf{Y}_T) \, = \, {\text {ln}} \, L({\varvec{\lambda }} | \mathbf{Y}_T)\) with respect to \({\varvec{\lambda }}\) leads to the score function \(S({\varvec{\lambda }})\).
We have
as
from the Bayes theorem. In the second line of the above computation, we have used the fact that \(Pr (\mathbf{z} | {\varvec{\lambda }}) \, = \, E(\mathbf{z} | {\varvec{\lambda }})\) is a constant function with respect to \({\varvec{\lambda }}\), hence its derivative vanishes.
Thus the score conditioned on a given regime vector \(\mathbf{z}\) has the expression
where
Since
we get the following relation
where
Deriving (20) with respect to \({\varvec{\theta }}_m\) gives
Substituting (21) into (19) yields formula (6) in Proposition 3.1.
Deriving (20) with respect to \(\mu _m\) and \(\pi _m\) produces
and
Substituting (22) resp. (23) into (19) gives formulae in (9). This completes the proof of Proposition 3.1.
Derivation of (10). From (8) and (9) we get
hence
Note that the expectation of \(\frac{\partial \, z_{t m}}{\partial \, \pi _m}\), conditional on \(\mathbf{Y}_{t - 1}\), vanishes. Summing up over \(m = 1, \dots , M\) produces
Thus relation (10) holds.
Derivation of (11). From (8) and (9) we have
hence
Note that the expectation of \(\frac{\partial \, z_{t m}}{\partial \, \mu _m}\), conditional on \(\mathbf{Y}_{t - 1}\), vanishes. Then we have
But \(y_t = {\mu }_m^{0} \, + \, \epsilon _t\) and \(\epsilon _t \, = \, \sqrt{h_{t m}^{0}}\, \eta _t\), where \({\mu }_m^0\) is the true value of the intercept and \(h_{tm}^0\) denote the function \(h_{t m}\) evaluated at the true value \({\varvec{\lambda }}_{m}^{0}\). Substituting this relation into the above expression of \({\hat{\mu }}_m \) yields (11). \(\square \)
Proof of Proposition 3.2
The consistency of \({\hat{\pi }}_m\) follows immediately from (10).
Using (11) and (12), the consistency of \({{\hat{\mu }}}_m \) follows from the sequence
as \(\eta _t\) is independent of \(s_t\) and \(h_{t m}\). Here C is a finite real constant. It remains to prove the consistency of \({\hat{\varvec{\theta }}}_m\) for all \(m = 1, \dots , M\). Using (6) and the FOC condition
we have
or, equivalently,
Note that the expectation of \(\frac{\partial \, z_{t m}}{\partial \, {\varvec{\theta }}_m}\), conditional on \(\mathbf{Y}_{t - 1}\), vanishes. Substituting \(y_t = \mu _{m}^0 \, + \, \epsilon _t = \mu _{m}^0 \, + \sqrt{h_{t m}^{0}} \, \eta _t\) into the last equation yields
hence
Taking the first Taylor expansion of \({{\hat{h}}}_{t m}\) around \(h_{t m}^{0}\), i.e.,
and substituting into the previous equation yields
Multiplying the last equation by \(T^{- 1}\) and taking the limit when T goes to infinity, we have
as \(E(\eta _t^2) = 1\) and \( {\text {plim}}_{T \rightarrow \infty } \, {{\hat{\mu }}}_m \, = \, {\mu }_m^0\), for all \(m = 1, \dots , M\). By assumption (13) the \(r \times r\) matrix \({{{\mathcal {I}}}}_m\) is finite and nonsingular. It follows that
which proves the consistency of \({\hat{\varvec{\theta }}}_m\). \(\square \)
Proof of Proposition 3.3
Set \({\varvec{\lambda }}_m^0 = {\varvec{\lambda }}_m\) to simplify notation. The asymptotic variance of \({{\hat{\pi }}}_m\) is given by
The aymptotic variance of \({{\hat{\mu }}}_m\) is given by
as \(E(\eta _t^2) = 1\) and \(E({{\hat{\tau }}}_{t m | t}^{ 2}) = E({\hat{\tau }}_{t m | t})\). To compute the asymptotic variance of \({\hat{\varvec{\theta }}}_{m}\) we set
and
Then we have
So it follows
Derivation of (14). The derivative \(\frac{\partial \, S_t ({\varvec{\theta }}_m)}{\partial \, {\varvec{\theta }}_{m}^{'}}\) is given by
Now we can approximate \( \frac{\partial ^2 \, {{{\mathcal {L}}}}({\varvec{\lambda }})}{\partial \, {\varvec{\theta }}_m \, \partial \, {\varvec{\theta }}_{m}^{'}}\) by taking the expectation of \(\frac{\partial \, S_t ({\varvec{\theta }}_m)}{\partial \, {\varvec{\theta }}_{m}^{'}}\) conditional on \(\mathbf{Y}_{t - 1}\). Recall that conditional on \(\mathbf{Y}_{t - 1}\) the magnitudes of \(h_{t m}\), \(\tau _{t m | t}\), and \( \frac{\partial \, \tau _{t m | t}}{ \partial \, {\varvec{\theta }}_{m}^{'}}\) are nonstochastic. Furthermore, we have
Then the conditional expectation of the second and third summand in \(\frac{\partial \, S_t ({\varvec{\theta }}_m)}{\partial \, {\varvec{\theta }}_{m}^{'}}\) vanish. So the conditional expectation of the first summand reduces to
Derivation of (15). The derivative \(\frac{\partial \, S_t ({\pi }_m)}{\partial \, {\pi }_{m}}\) is given by
Hence it follows
Derivation of (16). The derivative \(\frac{\partial \, S_t ({\mu }_m)}{\partial \, {\mu }_{m}}\) is given by
hence
Derivation of (18). The last term of
can be approximated by T/2 for T sufficiently large. In fact, we have
as \({{\hat{\epsilon }}}_t^2 \, = \, {{\hat{h}}}_{t m} \, \eta _t^2\) and \(\sum _{m = 1}^M \, {{\hat{\tau }}}_{t m | t} = 1\). For T large the term \(T^{- 1} \, \sum _{t = 1}^T \, \eta _t^2\) converges to \(E(\eta _t^2) = 1\), hence \( \frac{1}{2} \, \sum _{t = 1}^T \, \sum _{m = 1}^M \, \frac{{{\hat{\epsilon }}}_t^2}{{{\hat{h}}}_{t m}} \, {\hat{\tau }}_{t m | t} \) can be approximated by T/2 for T sufficiently large. By using (10), we get
for T sufficiently large. Then formula (18) follows.
Representing a MGARCH by a MS ARMA model. A MGARCH(M; p, q) model can be represented by a M-state Markov switching ARMA(r, p) process, where \(r = \max (p, q)\). Set \(\epsilon _t^2 = h_{t, s_t} + v_t\). Then we have
where \(\delta _{k, s_t} = \alpha _{k, s_t} \, + \, \beta _{k, s_t}\). Here we set \(\alpha _{k, s_t} = 0\) for \(k > q\) and \(\beta _{k, s_t} = 0\) for \(k > p\). For t sufficiently large, the disturbance process \(v_t\) can be approximately driven by a normal distribution with zero mean and variance
as \(E[(\eta _t^2 \, - \, 1)^2] = E(\eta _t^4) \, - \, 2\, E(\eta _t^2) \, + \, 1 \, = \, 2\) for \(\eta _t \sim N(0, 1)\). The explicit computation of unconditional moments of the powers of \(\epsilon _t^2\) and \(h_{t, s_t}\) can be found in Francq and Zakoïan (2008). \(\square \)
Rights and permissions
About this article
Cite this article
Cavicchioli, M. Statistical inference for mixture GARCH models with financial application. Comput Stat 36, 2615–2642 (2021). https://doi.org/10.1007/s00180-021-01092-5
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00180-021-01092-5