Linear quantile autoregressive model estimation using Stochastic EM algorithm

Document Type : Original Paper


Department of Statistics, Faculty of Basic Science, Zanjan University, Zanjan, Iran


In this paper, the quantile autoregressive time series model is introduce and then

the model parameters are estimated using the Stochastic EM algorithm, which is an iterative method

to compute maximum likelihood estimates. The likelihood function for the quantile autoregressive

model is constructed based on the asymmetric Laplace distribution and a scale mixture

representation of this distribution is used to estimate the model parameters via Stochastic EM algorithm.

The efficiency and application of the proposed method are illustrated by some simulation studies and analyzing a real dataset.


Main Subjects

[1] Barndorff O., Nielsen E., Non-Gassian Ornstein-Uhlenbeck Based Models some of their uses in financialEconomics, J. R. Stat. Soc., B: Stat. Methodol., 63 (2001), 167–241.
[2] Celeux G., Diebolt J., The SEM algorithm: A probabilistic teacher algorithm derived from the EMalgorithm for the mixture problem, Computational Statistics Quarterly, 2 (1985), 73–-82.
[3] Dempester A., Lird N., Rubin D., Maximum Likelihood from Incomplete Data via the EM algorithm,J. R. Stat. Soc., B: Stat. Methodol., 39 (1977), 1–38.
[4] Engle R.F., Manganelli S., CAViaR:Conditional autoregressive value at risk by regression quantiles,J. Bus. Econ. Stat., 22 (2004), 367–381.
[5] Ferreira G., Luis M.C., Lachos V.H., Bayesian modeling of autoregressive partial linear models withscale mixture of normal errors, J. Appl. Stat., 40 (2013), 1796–1816.
[6] Ghasemzadeh S., Ganjali M., Baghfalaki T., Quantile regression via the EM algorithm for joint modelingof mixed discrete and continuous data based on Gaussian copula, Stat. Methods Appl., 31 (2022),1–22.
[7] Hajrajabi A., Fallah A., Classical and Bayesian Estimate of the AR(1) Model with Skew Symmetric
innovations, J. Iran. Stat. Soc., 18 (2019), 157–175.
[8] Hajrajabi A., Maleki M., Nonlinear Semiparametric autoregressive model with finite mixtures of scalmixtures of skew normal innovations, J. Appl. Stat., 46 (2019), 2010–2029.
[9] Koenker R., Bassett J., Regression Quantiles, Econometrica, 46 (1978) 33–50.
[10] Koenker R., Xiao Z., Quantile autoregression, J. Am. Stat. Assoc., 101 (2006), 980–990.
[11] Kozumi H., Kobayashi G., Gibbs sampling methods of Bayesian quantile regression, J. Stat. Comput.Simul., 81 (2011), 1565–1578.
[12] Li G., Li Y., Tsai C.L., Quantile correlations and quantile autoregressive modeling, J. Am. Stat.
Assoc., 110 (2015), 246–261
[13] Liu J., Kumar S., Palomar D.P., Parameter estimation of Havy-tailed AR model with Missing Datavia Stochastic EM algorithm, IEEE Trans. Signal Process., 67 (2019), 2159–2172.
[14] Morán‐Vásquez R.A., Mazo‐Lopera M.A., Ferrari S.L., Quantile modeling through multivariate
log‐normal/independent linear regression models with application to newborn data, Biom. J., 63
(2021), 1290–1308.
[15] Nduka U.C., EM-based algorithms for autoregressive models with t-distributed innovations, Commun.Stat. - Simul. Comput., 47 (2018), 206–228.
[16] Tian Y., Tang M., Zang Y., Tian M., Quantile regression for linear models with autoregressive errorsusing EM algorithm, Comput. Stat., 33 (2018), 1605–1625.
[17] Tao Y., Yin J., Maximum likelihood estimation for quantile autoregression models with Markovianswitching, Commun. Stat. Theory Methods, (2022), 1–29.
[18] Yu K., Moyeed R.A., Bayesian quantile regression, Stat. Probab. Lett., 54 (2001), 437–447.
[19] Zhou Y.H., Ni Z.X., Li Y., Quantile Regression via the EM Algorithm, Commun. Stat. - Simul.
Comput., 43 (2014), 2162–2172.
[20] Zeng Z., Li M., Bayesian median autoregression for robust time series forecasting, Int. J. Forecast.,37 (2021), 1000–1010.