Econometric Analysis of Cross Section and Panel Data

1 Econometric Analysis of Cross Section and Panel DataJe rey M. WooldridgeThe MIT PressCambridge, MassachusettsLondon, EnglandContentsPrefacexviiAcknowledgment sxxiiiIINTRODUCTION AND Relationships and Ceteris Paribus Stochastic Setting and Asymptotic Data Asymptotic Not Fixed Explanatory Variables?92 Conditional Expectations and Related Concepts in Role of Conditional Expectations in of Conditional Definition and Partial E ects, Elasticities, and The Error Form of Models of Conditional Some Properties of Conditional Average Partial E Projections24 Problems27 Appendix Properties of Conditional Properties of Conditional Properties of Linear Projections323 Basic Asymptotic of Deterministic in Probability and Bounded in in Theorems for Random Behavior of Estimators and Test Asymptotic Properties of Asymptotic Properties of Test Statistics43 Problems45 IILINEAR MODELS474 The Single-Equation Linear Model and OLS of the Single-Equation Linear Properties of Asymptotic Inference Using heteroskedasticity - robust Lagrange

2 Multiplier (Score) Solutions to the Omitted Variables OLS Ignoring the Omitted The Proxy Variable OLS Models with Interactions in of OLS under Measurement Measurement Error in the Dependent Measurement Error in an Explanatory Variable73 Problems765 Instrumental Variables Estimation of Single-Equation Linear Variables and Two-Stage Least Motivation for Instrumental Variables Multiple Instruments: Two-Stage Least Treatment of Asymptotic Normality of Asymptotic E ciency of Hypothesis Testing with heteroskedasticity - robust Inference for Potential Pitfalls with Solutions to the Omitted Variables and Measurement Leaving the Omitted Factors in the Error Solutions Using Indicators of the Unobservables105 Problems1076 Additional Single-Equation with Generated Regressors and OLS with Generated 2 SLS with Generated Generated Instruments and Specification Testing for Testing Overidentifying Testing Functional Testing for Methods under Other Sampling Pooled Cross Sections

3 Over Geographically Stratified Spatial Cluster Samples134 Problems135 Appendix 6A1397 Estimating Systems of Equations by OLS and OLS Estimation of a Multivariate Linear Asymptotic Properties of System Testing Multiple and Asymptotic Normality of Generalized Asymptotic Asymptotic Asymptotic Variance of FGLS under a Using Unrelated Regressions, Comparison between OLS and FGLS for SUR Systems with Cross Equation Singular Variance Matrices in SUR Linear Panel Data Model, Assumptions for Pooled Dynamic A Note on Time Series robust Asymptotic Variance Testing for Serial Correlation and heteroskedasticity afterPooled Feasible GLS Estimation under Strict Exogeneity178 Problems1798 System Estimation by Instrumental and General Linear System of Method of Moments A General Weighting The System 2 SLS The Optimal Weighting The Three-Stage Least Squares Comparison between GMM 3 SLS and Traditional Considerations When Choosing an Using Testing Classical Testing Overidentification E cient Estimation and Optimal Instruments202 Problems2059 Simultaneous

4 Equations Scope of Simultaneous Equations in a Linear Exclusion Restrictions and Reduced General Linear Restrictions and Structural Unidentified, Just Identified, and Overidentified after The Robustness-E ciency Trade-o When Are 2 SLS and 3 SLS Equivalent? Estimating the Reduced Form Topics in Linear Using Cross Equation Restrictions to Achieve Using Covariance Restrictions to Achieve Subtleties Concerning Identification and E ciency in Nonlinear in Endogenous erent Instruments for Di erent Equations237 Problems23910 Basic Linear Unobserved E ects Panel Data Motivation: The Omitted Variables Assumptions about the Unobserved E ects and Random or Fixed E ects?

5 Strict Exogeneity Assumptions on the Some Examples of Unobserved E ects Panel Data Estimating Unobserved E ects Models by Pooled Random E ects Estimation and Inference under the Basic Random E robust Variance Matrix A General FGLS Testing for the Presence of an Unobserved E Fixed E ects Consistency of the Fixed E ects Asymptotic Inference with Fixed E The Dummy Variable Serial Correlation and the robust Variance Fixed E ects Using Fixed E ects Estimation for Policy First Di erencing robust Variance Testing for Serial Policy Analysis Using First Di Comparison of Fixed E ects versus First Di The Relationship between the Random E ects and FixedE ects The Hausman Test Comparing the RE and FE Estimators288 Problems29111 More Topics in Linear Unobserved E ects Unobserved E ects Models without the Strict Models under Sequential Moment Models with Strictly and Sequentially ExogenousExplanatory Models with Contemporaneous Correlation between SomeExplanatory Variables and the Idiosyncratic Summary of Models without Strictly ExogenousExplanatory Models with Individual-Specific A Random Trend General Models with Individual-Specific GMM Approaches to

6 Linear Unobserved E ects Equivalence between 3 SLS and Standard Panel Chamberlain s Approach to Unobserved E ects Hausman and Taylor-Type Applying Panel Data Methods to Matched Pairs and ClusterSamples328 Problems332 IIIGENERAL APPROACHES TO NONLINEAR Identification, Uniform Convergence, and Asymptotic Two-Step Asymptotic Estimating the Asymptotic Estimation without Nuisance Adjustments for Two-Step Hypothesis Wald Score (or Lagrange Multiplier) Tests Based on the Change in the Objective Behavior of the Statistics under Optimization The Newton-Raphson The Berndt, Hall, Hall.

7 And Hausman The Generalized Gauss-Newton Concentrating Parameters out of the Objective Simulation and Resampling Monte Carlo Bootstrapping378 Problems38013 Maximum Likelihood Preliminaries and General Framework for Conditional Consistency of Conditional Asymptotic Normality and Asymptotic Variance Asymptotic Estimating the Asymptotic Hypothesis Specification Partial Likelihood Methods for Panel Data and Cluster Setup for Panel Asymptotic Inference with Dynamically Complete Inference under Cluster Panel Data Models with Unobserved E Models with Strictly Exogenous Explanatory Models with Lagged Dependent Two-Step MLE413 Problems414 Appendix 13A41814 Generalized Method of Moments and Minimum Distance Asymptotic Properties of Estimation under Orthogonality Systems of Nonlinear Panel Data E cient A General E ciency E ciency of E cient Choice of Instruments under Conditional Classical Minimum Distance Estimation442 Problems446 Appendix 14A448 IVNONLINEAR MODELS AND RELATED TOPICS45115 Discrete Response The Linear Probability Model for Binary Index Models for Binary Response.

8 Probit and Maximum Likelihood Estimation of Binary Response Testing in Binary Response Index Testing Multiple Exclusion Testing Nonlinear Hypotheses Tests against More General Reporting the Results for Probit and Specification Issues in Binary Response Neglected Continuous Endogenous Explanatory A Binary Endogenous Explanatory heteroskedasticity and Nonnormality in the LatentVariable Estimation under Weaker Binary Response Models for Panel Data and Cluster Pooled Probit and Unobserved E ects Probit Models under Strict Unobserved E ects Logit Models under Strict Dynamic Unobserved E ects Semiparametric Cluster Multinomial Response Multinomial Probabilistic Choice Ordered Response Ordered Logit and Ordered Applying Ordered Probit to Interval-Coded Data508 Problems50916 Corner Solution Outcomes and Censored Regression Introduction and Derivations of Expected Inconsistency of Estimation and Inference with Censored Reporting the Specification Issues in Tobit Neglected Endogenous Explanatory heteroskedasticity and Nonnormality in the LatentVariable Estimation under Conditional Median Some Alternatives to Censored Tobit for Corner

9 Applying Censored Regression to Panel Data and Cluster Pooled Unobserved E ects Tobit Models under Strict Dynamic Unobserved E ects Tobit Models542 Problems54417 Sample Selection, Attrition, and Stratified When Can Sample Selection Be Ignored? Linear Models: OLS and Nonlinear Selection on the Basis of the Response Variable: A Probit Selection Exogenous Explanatory Endogenous Explanatory Binary Response Model with Sample A Tobit Selection Exogenous Explanatory Endogenous Explanatory Estimating Structural Tobit Equations with Sample Sample Selection and Attrition in Linear Panel Data Fixed E ects Estimation with Unbalanced Testing and Correcting for Sample Selection Stratified Standard Stratified Sampling and Variable Weighted Estimators to Account for Stratification Based on Exogenous Variables596 Problems59818 Estimating Average Treatment E A Counterfactual Setting and the

10 Self-Selection Methods Assuming Ignorability of Regression Methods Based on the Propensity Instrumental Vari