Multi-dimensional Spatio-temporal and Network Modelling
Saturday, Jan. 5, 2019 12:30 PM - 2:15 PM
- Chair: Yongcheol Shin, University of York
A Time-Space Dynamic Panel Data Model with Spatial Moving Average Errors
AbstractThis paper focuses on the estimation and predictive performance of several estimators for the time-space dynamic panel data model with Spatial Moving Average Random Effects (SMA-RE) structure of the disturbances. A dynamic spatial Generalized Moments (GM) estimator is proposed which combines the approaches proposed by Baltagi, Fingleton and Pirotte (2014) and Fingleton (2008). The main idea is to mix non-spatial and spatial instruments to obtain consistent estimates of the parameters. Then, a forecasting approach is proposed and a linear predictor is derived. Using Monte Carlo simulations, we compare the short-run and long-run effects and evaluate the predictive efficiencies of optimal and various suboptimal predictors using the Root Mean Square Error (RMSE) criterion. Last, our approach is illustrated by an application in geographical economics which studies the employment levels across 255 NUTS regions of the EU over the period 2001-2012, with the last two years reserved for prediction.
Normal Approximation in Dynamic Network Formation
AbstractWe prove a general central limit theorem for network statistics satisfying a high-level "stabilization" condition. The condition essentially requires that the contribution of each node to the statistic only depends on a small number of nodes in the network and therefore can be interpreted as a weak-dependence condition. We apply this result to dynamic models of network formation with strategic interactions and derive primitive sufficient conditions under which important classes of network moments satisfy stabilization.
Limit Theorems for Data with Network Structure
AbstractThis paper considers data generated by endogenous networks. Networks are formed strategically. The probability of forming a link between individuals is based on observable and unobservable characteristics. Outcomes of interest depend on the structure of the network. Endogeneity arises from unobservable link characteristics that may be correlated with outcome variables. Observable characteristics include location in physical or attribute space and lead to homophily in link formation. The paper imposes primitive conditions on the distribution of observable characteristics, unobservables and functional forms determining link formation. A summability condition over the probability distribution of observable network locations is shown to be a critical ingredient in establishing laws of large numbers for a class of network moments. Convergence of these moments are high level conditions for GMM estimators developed in Kuersteiner and Prucha (2013, 2015) as well as the central limit theorems in these papers. The current paper then provides a set of sufficient low level conditions for these central limit theorems.
The Spatio-Temporal Autoregressive Distributed Lag Modelling Approach to an Analysis of the Spatial Heterogeneity and Diffusion Dependence
AbstractGiven the growing availability of large datasets and following recent research trends on multi-dimensional modelling, we propose the novel spatio-temporal autoregressive distributed lag (STARDL) models and analyses the general case where spatial and temporal coefficients differ jointly across the spatial units. To date, the parameters of spatial panel data model have been assumed to be homogeneous across units, with Aquaro, Bailey and Pesaran (2015) the only paper in considering heterogeneous spatial parameters but without any dynamics. Our model encompasses theirs, while also encompassing the widely used spatial dynamic panel data model. To deal with a degree of simultaneity associated with the spatial-lagged dependent variables, we develop a control function-based STARDL estimator by employing time- and spatial-lagged regressors as a variety of instruments. In particular, we aim to develop the asymptotic theory for the STARDL estimators of individual spatial and temporal coefficients and establish that they are consistent and asymptotically normally distributed when both the time and cross section dimensions of the panel are large. An important feature of the STARDL model is to capture three different forms of dynamic adjustment from initial equilibrium to the new equilibrium following an economic perturbation with respect to three different types of regressors. In this regard we derive the generalised dynamic spatial panel data model representation from which we develop the dynamic, the spatial and the diffusion multipliers. As the estimates are allowed to be individual-specific, the analysis of the dynamic or diffusion multipliers can be further conducted through network/graph approach. The salient features of the proposed model are illustrated by the empirical applications to the Iraqi war casualties during 2003-2010.
- C1 - Econometric and Statistical Methods and Methodology: General
- C3 - Multiple or Simultaneous Equation Models; Multiple Variables