An estimation of minimum second powers in Arch model parameters in the presence of missing data

Document Type : Research Paper

Authors

Allameh Tabataba'i University

Abstract

Many financial and economic time series exhibit high volatility in certain periods and have relatively less clustered volatility in others, indicating cluster-wise volatility. Assuming constant variance in such cases is not reasonable. One of the modeling tools for these variations is the use of Autoregressive Conditional Heteroscedastic (ARCH) and Generalized Autoregressive Conditional Heteroscedastic (GARCH) models. The ARCH model has various applications in econometrics and finance, including the study of volatility changes related to stock indices, gold prices, and exchange rates. One of the common methods for estimating ARCH model parameters is the Quasi-Maximum Likelihood Estimator (QMLE) method. As this estimator does not have a closed form, the estimator of the Minimum Second Powers is used for estimating ARCH model parameters, which is more efficient compared to other estimators. Missing data in time series are common, and therefore, we obtain the two-stage Minimum Second Powers estimator in the presence of missing data. This estimator has the same asymptotic efficiency as the Quasi-Maximum Likelihood Estimator. We prove its strong consistency and asymptotic normality and confirm these results through simulations. The results are applied to the data of the Tehran Stock Exchange overall index.

Keywords

Main Subjects

References

[1] R. Fletcher and S. Leyffer, Nonlinear programming without a penalty function, Math. Program., 91 no. 2 (2002) 239–269.
[2] A. Bose and K. Mukherjee, Estimating the arch parameters by solving linear equations, J. Time Ser. Anal., 24 no. 2 (2003) 127–136.
[3] E. Parzen, On spectral analysis with missing observations and amplitude modulation, Sankhyā Ser. A, 25 (1963) 383–392.
[4] N. Shephard, Statistical aspect of ARCH and stochastic volatility, In D. R. Cox, D. V. Hinkley, O. E. Barndorff-Nielsen (Eds.), Time series models in econometrics, finance and other fields, London: Chapman and Hall, 1996.
[5] A. A. Weiss, Asymptotic theory for ARCH models: Estimation and testing, Econometric Theory, 2 no. 1 (1986) 107–131.
Mathematics and Society
Volume 2, Issue 4 - Serial Number 4
December 2017
Pages 1-10

Files

History

  • Receive Date: 01 October 2014
  • Revise Date: 04 January 2017
  • Accept Date: 19 August 2017
  • Publish Date: 20 February 2018

Share

How to cite

Statistics