Three essays in spatial econometrics

Sen, Monalisa

Permalink

https://hdl.handle.net/2142/34272 Copy

Description

Title

Three essays in spatial econometrics

Author(s)

Sen, Monalisa

Issue Date

2012-09-18T21:08:54Z

Director of Research (if dissertation) or Advisor (if thesis)

Bera, Anil K.

Doctoral Committee Chair(s)

Bera, Anil K.

Committee Member(s)

• Hewings, Geoffrey J.D.
• McMillen, Daniel P.
• Shao, Xiaofeng

Department of Study

Economics

Discipline

Economics

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Spatial Econometrics
• Specification Tests
• Rao's Score
• Growth Convergence
• Spatial Auto-regressive Models

Abstract

A focus on location and spatial interaction has recently gained a more central place not only in applied econometrics but also in theoretical econometrics. The standard econometric techniques often fail in the presence of spatial autocorrelation, which is commonplace in geographic (cross-sectional) data sets, and thus giving misleading inference, and thereby wrong policy implications are derived from these econometric models. In this dissertation I have dealt with such model specification issues which arises due to spatial nature of the data. The main contributions of this dissertation are thus two folds: spatial model specification and specification tests for spatial panel data models. Each chapter provides econometric methods along with empirical examples to demonstrate the importance and utility of the proposed methods. In Chapter 2, I propose an estimation strategy for popular spatial weight matrix. The spatial lag dependence in a regression model is similar to the inclusion of a serially autoregressive term for the dependent variable in time- series context. However, unlike in the time series model, the implied covariance structure matrix from the spatial autoregressive model can have a very counterintuitive and improbable structure. However, if the weight matrix can capture the underlying dependence structure of the observations then this unintuitive behavior of implied correlation gets corrected to a large extent. Thus in Chapter 2, I explore the possibility of constructing the weight matrix (or the overall spatial dependence in the data) that is consistent with the underlying correlation structure of the dependent variable. Specification of a model is one of the most fundamental problems in econometrics. However, in most cases, specification tests are carried out in a piecemeal fashion, for example, testing the presence of one-effect at a time ignoring the potential presence of other forms of misspecification. In Chapter 3, I overcome these difficulties by proposing adjusted RS tests for the panel spatial models under a very general framework, and my proposed test statistics are robust under multiple-forms of misspecification. Most of the existing procedures like likelihood ratio tests (LR) and conditional Lagrange multiplier (LM) tests require estimation of the nuisance parameters. In this respect, a very attractive feature of my approach is that the adjusted tests are based on the joint null hypothesis (of no misspecification) as each the proposed tests take care of the possible presence of all the nuisance parameters through their respective Fisher-Rao score evaluated under joint null and thus requiring estimation of the simplest model. In Chapter 4, I develop on the theoretical foundation of Chapter 3, by proposing the size-robust tests for dynamic panel models with dynamic space-time dependence. I propose the test statistics robust under local misspecification for time dynamics, individual effects, serial correlation of errors and spatial dependencies like spatial lag and error, and time-space dynamics under the dynamic panel model. Using these proposed tests I investigate the salient features of the data that truly matters for growth analyses. In growth theory different kinds of econometric models have been proposed based on economic theory and the subjective beliefs of researchers, - including simple cross-sectional regression models, panel data models, time series models and recently many types of spatial models. Unfortunately the estimate of growth convergence rate under these different model frameworks vary wildly, even when the same dataset is used. Thus, the question becomes: which model is most appropriate? I propose to address this problem by developing six adjusted Rao's score (RS) tests that are robust under misspecification for a very general dynamic panel model. I start with a simple panel model and then using my proposed test statistics I check whether particular departures (like time dynamics, serial correlation, individual effects, spatial/cross sectional dependence) from this initial specification are supported or rejected by the data. Thus Chapter 4 contributes both to the econometric methodology of specification tests and also tackles with the empirical question of growth convergence debate.

Graduation Semester

2012-08

Permalink

http://hdl.handle.net/2142/34272

Copyright and License Information

Copyright 2012 Monalisa Sen

Owning Collections