RT Dissertation/Thesis
T1 Modelling of spatial effects in count data
A1 Glaser,Stephanie
WP 2017/04/06
AB In this thesis, spatial structures in discrete valued count observations are modelled. More precisely, a global spatial autocorrelation parameter is estimated in the framework of a nonlinear count data regression model. For this purpose, cross-sectional and panel count data models are developed which incorporate spatial autocorrelation and allow for additional explanatory variables. The proposed models include the so-called "Spatial linear feedback model" for cross-sectional data as well as for panel data including fixed effects, which is estimated using maximum likelihood estimation. Additionally, two approaches for a distribution-free panel estimation using GMM are presented. The models are applied to a cross-sectional U.S. start-up firm births data set and a panel data set with crime counts from Pittsburgh.
K1 Ökonometrie
K1 Zähldaten
K1 Räumliche Statistik
K1 Unternehmensgründung
K1 Kriminalstatistik
PP Hohenheim
PB Kommunikations-, Informations- und Medienzentrum der Universität Hohenheim
UL http://opus.uni-hohenheim.de/volltexte/2017/1345