On the existence of the maximum likelihood estimates in Poisson regression

被引:223
作者
Silva, J. M. C. Santos [1 ,2 ]
Tenreyro, Silvana [3 ,4 ]
机构
[1] Univ Essex, Colchester CO4 3SQ, Essex, England
[2] CEMAPRE, Colchester CO4 3SQ, Essex, England
[3] London Sch Econ, CREI, CEP, London WC2A 2AE, England
[4] CEPR, Dept Econ, London WC2A 2AE, England
关键词
Collinearity; Complete separation; Perfect prediction; MODELS; LOG;
D O I
10.1016/j.econlet.2010.02.020
中图分类号
F [经济];
学科分类号
02 ;
摘要
We note that the existence of the maximum likelihood estimates in Poisson regression depends on the data configuration, and propose a strategy to identify the existence of the problem and to single out the regressors causing it. (C) 2010 Elsevier B.V. All rights reserved.
引用
收藏
页码:310 / 312
页数:3
相关论文
共 8 条
[1]  
ALBERT A, 1984, BIOMETRIKA, V71, P1
[2]  
[Anonymous], 2013, Regression Analysis of Count Data
[3]  
[Anonymous], 2008, Econometric analysis of count data, DOI DOI 10.1007/978-3-540-24728-9_5
[4]   PSEUDO MAXIMUM-LIKELIHOOD METHODS - APPLICATIONS TO POISSON MODELS [J].
GOURIEROUX, C ;
MONFORT, A ;
TROGNON, A .
ECONOMETRICA, 1984, 52 (03) :701-720
[5]   LOG-LINEAR MODELS FOR FREQUENCY DATA - SUFFICIENT STATISTICS AND LIKELIHOOD EQUATIONS [J].
HABERMAN, SJ .
ANNALS OF STATISTICS, 1973, 1 (04) :617-632
[6]   Estimating log models: to transform or not to transform? [J].
Manning, WG ;
Mullahy, J .
JOURNAL OF HEALTH ECONOMICS, 2001, 20 (04) :461-494
[7]  
SANTNER TJ, 1986, BIOMETRIKA, V73, P755
[8]   The log of gravity [J].
Silva, J. M. C. Santos ;
Tenreyro, Silvana .
REVIEW OF ECONOMICS AND STATISTICS, 2006, 88 (04) :641-658