One-step robust estimation of fixed-effects panel data models

M. Aquaro, P. Cizek

Research output: Contribution to journalArticleScientificpeer-review

Abstract

The panel-data regression models are frequently applied to micro-level data, which often suffer from data contamination, erroneous observations, or unobserved heterogeneity. Despite the adverse effects of outliers on classical estimation methods, there are only a few robust estimation methods available for fixed-effects panel data. A new estimation approach based on two different data transformations is therefore proposed. Considering several robust estimation methods applied to the transformed data, the robust and asymptotic properties of the proposed estimators are derived, including their breakdown points and asymptotic distributions. The finite-sample performance of the existing and proposed methods is compared by means of Monte Carlo simulations.
Original languageEnglish
Pages (from-to)536-548
JournalComputational Statistics & Data Analysis
Volume57
Issue number1
DOIs
Publication statusPublished - 2013

Fingerprint

Fixed Effects
Robust Estimation
Panel Data
Data Model
Data structures
Unobserved Heterogeneity
Breakdown Point
Data Transformation
Contamination
Asymptotic distribution
Asymptotic Properties
Outlier
Regression Model
Monte Carlo Simulation
Estimator

Cite this

@article{fc7c838472cd4fd7a730e05a5308e17c,
title = "One-step robust estimation of fixed-effects panel data models",
abstract = "The panel-data regression models are frequently applied to micro-level data, which often suffer from data contamination, erroneous observations, or unobserved heterogeneity. Despite the adverse effects of outliers on classical estimation methods, there are only a few robust estimation methods available for fixed-effects panel data. A new estimation approach based on two different data transformations is therefore proposed. Considering several robust estimation methods applied to the transformed data, the robust and asymptotic properties of the proposed estimators are derived, including their breakdown points and asymptotic distributions. The finite-sample performance of the existing and proposed methods is compared by means of Monte Carlo simulations.",
author = "M. Aquaro and P. Cizek",
note = "Appeared earlier as CentER Discussion Paper 2010-110",
year = "2013",
doi = "10.1016/j.csda.2012.07.003",
language = "English",
volume = "57",
pages = "536--548",
journal = "Computational Statistics & Data Analysis",
issn = "0167-9473",
publisher = "Elsevier",
number = "1",

}

One-step robust estimation of fixed-effects panel data models. / Aquaro, M.; Cizek, P.

In: Computational Statistics & Data Analysis, Vol. 57, No. 1, 2013, p. 536-548.

Research output: Contribution to journalArticleScientificpeer-review

TY - JOUR

T1 - One-step robust estimation of fixed-effects panel data models

AU - Aquaro, M.

AU - Cizek, P.

N1 - Appeared earlier as CentER Discussion Paper 2010-110

PY - 2013

Y1 - 2013

N2 - The panel-data regression models are frequently applied to micro-level data, which often suffer from data contamination, erroneous observations, or unobserved heterogeneity. Despite the adverse effects of outliers on classical estimation methods, there are only a few robust estimation methods available for fixed-effects panel data. A new estimation approach based on two different data transformations is therefore proposed. Considering several robust estimation methods applied to the transformed data, the robust and asymptotic properties of the proposed estimators are derived, including their breakdown points and asymptotic distributions. The finite-sample performance of the existing and proposed methods is compared by means of Monte Carlo simulations.

AB - The panel-data regression models are frequently applied to micro-level data, which often suffer from data contamination, erroneous observations, or unobserved heterogeneity. Despite the adverse effects of outliers on classical estimation methods, there are only a few robust estimation methods available for fixed-effects panel data. A new estimation approach based on two different data transformations is therefore proposed. Considering several robust estimation methods applied to the transformed data, the robust and asymptotic properties of the proposed estimators are derived, including their breakdown points and asymptotic distributions. The finite-sample performance of the existing and proposed methods is compared by means of Monte Carlo simulations.

U2 - 10.1016/j.csda.2012.07.003

DO - 10.1016/j.csda.2012.07.003

M3 - Article

VL - 57

SP - 536

EP - 548

JO - Computational Statistics & Data Analysis

JF - Computational Statistics & Data Analysis

SN - 0167-9473

IS - 1

ER -