Likelihood approximation by numerical integration on sparse grids

被引：262

作者：

Heiss, Florian ^{[1
]}

Winschel, Viktor ^{[2
]}

机构：

[1] Univ Munich, Dept Econ, D-80539 Munich, Germany

[2] Univ Mannheim, Dept Econ, D-68163 Mannheim, Germany

来源：

JOURNAL OF ECONOMETRICS | 2008年 / 144卷 / 01期

关键词：

likelihood simulation; multivariate quadrature; mixed logit;

D O I：

10.1016/j.jeconom.2007.12.004

中图分类号：

F [经济];

学科分类号：

02 ;

摘要：

The calculation of likelihood functions of many econometric models requires the evaluation of integrals without analytical solutions. Approaches for extending Gaussian quadrature to multiple dimensions discussed in the literature are either very specific or suffer from exponentially rising computational costs in the number of dimensions. We propose an extension that is very general and easily implemented, and does not suffer from the curse of dimensionality. Monte Carlo experiments for the mixed logit model indicate the superior performance of the proposed method over simulation techniques. (C) 2008 Elsevier B.V. All rights reserved.

引用

页码：62 / 80

页数：19

共 38 条

[1] Quasi-random maximum simulated likelihood estimation of the mixed multinomial logit model [J].