Neighbourhood GMM estimation of dynamic panel data models

Vasilis Sarafidis

Research output: Contribution to journalArticleResearchpeer-review

8 Citations (Scopus)


A new approach is developed for estimation of short dynamic panel data models with spatially correlated errors. The method employs an additional set of moment conditions that become available for each i—specifically, instruments with respect to the individual(s) which unit i is spatially correlated with. These moment conditions are non-redundant and remain informative even if the data generating process is close to a unit root one. The proposed GMM estimator is consistent and asymptotically normally distributed. An extensive Monte Carlo study also builds a GMM estimator that combines spatial and standard instruments. This estimator appears to perform very well under a wide range of parametrisations in terms of both bias and root mean square error. The proposed method is illustrated using crime data based on a panel of 153 local government areas in NSW, spanning a period of 5 years.

Original languageEnglish
Pages (from-to)526-544
Number of pages19
JournalComputational Statistics and Data Analysis
Publication statusPublished - 29 May 2016


  • Dynamic panel data
  • Generalised Method of Moments
  • Spatial dependence

Cite this