All Rights Reserved
AccessEcon LLC 2006, 2008.
Powered by MinhViet JSC

 
Zsolt Sándor
 
''Further evidence on sparse grids-based numerical integration in the mixed logit model''
( 2019, Vol. 39 No.4 )
 
 
We study the performance of Gauss quadrature methods based on sparse grids for approximating integrals involved in mixed logit models. In Monte Carlo experiments we consider data generating processes in which consumer heterogeneity has low variance and data generating processes in which it has high variance. In the former case we find that, in line with previous literature, sparse grids produce very accurate estimates even when the number of points used for approximating integrals is small. However, in the latter case sparse grids yield biased estimates and are outperformed by quasi-Monte Carlo methods.
 
 
Keywords: discrete choice; random coefficients; simulation; quasi-Monte Carlo; panel
JEL: C1 - Econometric and Statistical Methods: General
C6 - Mathematical Methods and Programming: General
 
Manuscript Received : Jul 25 2019 Manuscript Accepted : Nov 29 2019

  This abstract has been downloaded 978 times                The Full PDF of this paper has been downloaded 166319 times