Linear Pseudo Panel Data Models with Multifactor Error Structure
Arturas Juodis (University of Amsterdam)
When genuine panel data samples are not available, repeated cross-section surveys can be used to form so-called pseudo panels. In this paper we investigate the properties of linear pseudo panel data models with fixed number of cohorts and time observations. As a base for our setup we take the study of Inoue (2008), extend it to models with unbalanced samples and show that the FE estimator is invariant to the filtering method used. Furthermore, we enlarge the original setup to models with cross-sectional dependence of factor type with unbalanced samples by adapting the approach of Ahn et al. (2013) for genuine panels. We discuss testing, identification and model selection issues. Finally, an extensive Monte Carlo simulation study is conducted to investigate the robustness of available estimation techniques with respect to endogeneity, cross-sectional dependence and near non-identification.
A comparison of test procedures in dynamic panel data models under weak identification.
Rutger Poldermans (University of Amsterdam)
We consider the linear dynamic panel data model with an additional endogenous regressor, where we focus on the accuracy of various Wald and weak identification robust GMM statistics. The latter, although robust to identification strength, seem vulnerable to the use of many moment conditions. However, implementations exploiting particular subsets of a limited number of instruments are more or less size correct. Additionally, using uncentered moments in robust covariance matrix estimation improves behavior under the null hypothesis . When centered moments are used in robust covariance matrix estimation the size cannot be controlled for.
Exogeneity Tests in Dynamic Panel Data Models
Milan Pleus (University of Amsterdam)
Exogeneity tests are investigated in the context of linear dynamic panel data models, estimated by GMM or CUE. Rather than testing all overidentifying restrictions by the Sargan-Hansen test, the focus is on classifying explanatory variables using either the incremental Sargan-Hansen test or the Durbin-Wu-Hausman test. As misclassification yields either inconsistent or inefficient estimates, testing the exogeneity status of variables is crucial. Although it is known in the literature that the Sargan-Hansen test suffers when using many instruments, it is yet unclear in what way the incremental test is affected. Additionally, test statistics are considered in which the number of employed instruments is deliberately restricted and the procedure of Hayakawa (2014) is generalized to the incremental test. With respect to the Hausman test a version with Windmeijer corrected variance is considered. Simulation is used to investigate finite sample performance.