Approximate estimation in nonlinear panel data models

Offer Lieberman, László Mátyás

Research output: Contribution to journalArticlepeer-review

1 Scopus citations


This paper is concerned with the estimation of a general class of nonlinear panel data models in which the conditional distribution of the dependent variable and the distribution of the heterogeneity factors are arbitrary In general, exact analytical results for this problem do not exist. Here, Laplace and small-sigma appriximations for the marginal likelihood are presented. The computation of the MLE from both approximations is straightforward. It is shown that the accuracy of the Laplace approximation depends on both the sample size and the variance of the individual effects, whereas the accuracy of the small-sigma approximation is O(1) with respect to the sample size. The results are applied to count, duration and probit panel data models. The accuracy of the approximations is evaluated through a Monte Carlo simulation experiment. The approximations are also applied in an analysis of youth unemployment in Australia.

Original languageEnglish
Pages (from-to)1177-1195
Number of pages19
JournalCommunications in Statistics Part B: Simulation and Computation
Issue number3
StatePublished - 1997
Externally publishedYes

Bibliographical note

Funding Information:
Computing assistance by Mark N. Harris and financial support by the Australian Research Council are kindly acknowledged.


  • Count data model
  • Duration model
  • Laplace approximation
  • Probit model
  • Small-sigma approximation
  • Solomon-Cox method


Dive into the research topics of 'Approximate estimation in nonlinear panel data models'. Together they form a unique fingerprint.

Cite this