REGRESSION WITH VOLATILE ERRORS IN THE PRESENCE OF MEASUREMENT ERRORS

Title

REGRESSION WITH VOLATILE ERRORS IN THE PRESENCE OF MEASUREMENT ERRORS

Creator

Thomas, Anna; John, Nimitha

Description

This study explores the estimation and testing of regression models with volatile errors when measurement errors are present. The presence of measurement error in models with heteroscedastic disturbances, such as those following an autoregressive conditional heteroscedasticity (ARCH) or Generalized ARCH (GARCH) structure, can lead to biased estimates and misleading inferences. To address this, we develop an estimation framework that accounts for both heteroscedasticity and mismeasured observations, ensuring consistent and asymptotically normal parameter estimates. We estimate the parameters using corrected score estimation and weighted linear regression, which effectively mitigate the impact of measurement error and hetroscedasticity. Additionally, we perform a Likelihood Ratio (LR) test to assess the significance of measurement errors in regression models with volatile errors. Through Monte Carlo simulations, we analyze the bias and efficiency of traditional estimators and demonstrate the robustness of our proposed approach. Finally, the methodology is applied to real-life economic and financial data, illustrating its practical relevance and effectiveness in empirical research. The findings contribute to improving statistical inference in models where measurement error and volatility coexist, ensuring more reliable and accurate parameter estimation. 2025, Gnedenko Forum. All rights reserved.

Source

Reliability: Theory and Applications;Volume;20;Issue;3;pp.310-321

Date

01-01-2025

Publisher

Gnedenko Forum

Subject

corrected score; likelihood ratio test; measurement error; regression; volatile errors

Coverage

Thomas A., Department of Statistics and Data Science CHRIST(Deemed to be University), Karnataka, Bengaluru, India; John N., Department of Statistics and Data Science CHRIST(Deemed to be University), Karnataka, Bengaluru, India

Rights

Restricted Access; Hardcopy may be available in the library

Relation

ISSN: 19322321;

Format

online

Language

English

Type

Article

Identifier

https://doi.org/10.24412/1932-2321-2025-386-310-321

https://www.scopus.com/pages/publications/105018068486?origin=resultslist