InferÃncia em modelos heteroscedÃsticos na presenÃa de pontos de alavanca

AUTOR(ES)

Tatiene Correia de Souza

DATA DE PUBLICAÇÃO

2003

RESUMO

The chief goal of this thesis is to study the finite-sample behavior of different heteroskedasticity- consistent covariance matrix estimators, under both constant and unequal error variances. We consider the estimator proposed by Halbert White (HC0), its variant known as HC3, and Wuâs (1986) weighted bootstrap estimator. Recently proposed estimators, such as CribariâNetoâs (2004) HC4 and CribariâNeto and Zarkosâs (2004) inversely adjusted weighted bootstrap, are also considered. We propose a new covariance matrix estimator: HC5. It is the first consistent estimator to explicitly take into account the effect that the maximum level of leverage of the data has on the associated inference. Our numerical (Monte Carlo) results show that quasi-t inference based on HC5 is typically more reliable than inference based on other covariance matrix estimators. We also present four applications to real data

ASSUNTO(S)

pontos de alavanca modelos heteroscedÃsticos inferÃncia estatistica

Documentos Relacionados

• Testes exatos em modelos heteroscedÃsticos
• CorreÃÃo de Bartlett nos modelos nÃo-lineares simÃtricos heteroscedÃsticos
• InferÃncia baseada em intervalos estatÃsticos aplicada à SeguranÃa PÃblica
• EstimaÃÃo dos parÃmetros de um cÃrculo para modelos heteroscedÃsticos de regressÃo
• Uma anÃlise da QoS na transmissÃo em redes 802.11G sob presenÃa de interferÃncia Bluetooth