************** Please note change of address *********************
***** Our department has moved to new premises. ***************
***** A map with instructions on how to get to our **************
***** department can be found at the following web site: *********
***** http://www.ucl.ac.uk/Stats/map.html ************************
Everyone is welcome. Tea, coffee and biscuits are provided!
Details of the seminar:
================
Monday 31 January 2000 - 4pm, Room 102, 1-19 Torrington Place,
Department of Statistical Science - University College London.
Speaker : Dr V. A. Hadjivassiliou
Title: Some Practical Issues in Maximum Simulated Likelihood
Abstract: In this paper I explore ways of recapturing the efficiency
property for
estimators that rely on simulation. In particular, I show that this can be
achieved by exploiting two-step maximum simulated likelihood (MSL) estimation
methods that are familiar from classical applications. I also construct
a diagnostic test for adequacy of number of simulations employed to
guarantee negligible bias for the MSL and provide some evidence on the
computational requirements of the Geweke-Hajivassiliou-Keane (GHK)
simulator as a function of (a) the dimension of the problem and (b) the
number of simulations employed in a vectorized context. I outline how one
can derive a similar approach for checking the adequacy of the number of
Gibbs resamplings in simulation estimation methods that employ this technique.
This paper also shows how to suitably introduce simulation into classical
hypothesis testing methods and provide test statistics (simulated Wald,
Lagrange Multiplier, and Likelihood Ratio Tests) that are free of
influential simulation noise.
Finally, I explain how simulation-variance-reduction techniques, notably
antithetics, can improve substantially the practical performance of the GHK
simulator and present extensive Monte-Carlo evidence confirming this.
----------------------------------------------------------------------------
----------
K. Skouras, Lecturer in Statistics, Department of Statistical Science, UCL,
1-19 Torringhton Place, London WC1E 6BT, United Kingdom.
Tel: 020 - 76791862 Fax: 0171-7383 4703
e-mail: [log in to unmask]
----------------------------------------------------------------------------
-----------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|