Advanced Studies in Theoretical and Applied Econometrics
Vo l ume 4 6
Managing Editors:
J. Marquez, The Federal Reserve Board, Washington, D.C., U.S.A
A. Spanos, Virginia Polytechnic Institute and State University, Blacksburg, VA, U.S.A
Editorial Board:
F.G. Adams, University of Pennsylvania, Philadelphia, U.S.A
P. Ba l e s t ra?, University of Geneva, Switzerland
M.G. Dagenais, University of Montreal, Canada
D. Kendrick, University of Texas, Austin, U.S.A
J.H.P. Paelinck, Netherlands Economic Institute, Rotterdam, The Netherlands
R.S. Pindyck, Sloane School of Management, M.I.T., U.S.A
W. Wel fe, University of Lodz, Poland

Contents
Part I Fundamentals
1 Introduction ............................................... 3
Marc Nerlove, Patrick Sevestre and Pietro Balestra
1.1 Introduction . . ............................................ 3
1.2 Data, Data-Generating Processes (DGP), and Inference . ........ 4
1.3 History and Dynamics ..................................... 8
1.4 A Brief Review of Other Methodological Developments . ........ 13
1.5 Conclusion . . . ............................................ 21
References . . . . . ................................................ 21
2 Fixed Effects Models and Fixed Coef?cients Models .............. 23
Pietro Balestra and Jayalakshmi Krishnakumar
2.1 The Covariance Model: Individual Effects Only . ............... 24
2.1.1 Speci?cation ..................................... 24
2.1.2 Estimation ....................................... 25
2.1.3 Inference . . . . ..................................... 28
2.2 The Covariance Model: Individual and Time Effects . . . . ........ 29
2.2.1 TimeEffectsOnly................................. 29
2.2.2 Time and Individual Effects . . . . . . ................... 30
2.2.3 Inference . . . . ..................................... 32
2.3 Non-spherical Disturbances . . . . . . . .......................... 33
2.3.1 What Variance–Covariance Stucture? . . ............... 33
2.3.2 Two General Propositions for Fixed Effects Models ..... 34
2.3.3 Individual Fixed Effects and Serial Correlation ......... 36
2.3.4 Heteroscedasticity in Fixed Effects Models . . . . ........ 38
2.4 Extensions............................................... 40
2.4.1 ConstantVariablesinOneDimension................. 40
2.4.2 VariableSlopeCoef?cients ......................... 41
2.4.3 Unbalanced Panels . . . . . . .......................... 44
References . . . . . ................................................ 48
viiviii Contents
3 Error Components Models ................................... 49
Badi H. Baltagi, L′ aszl ′ oM′ aty′ as and Patrick Sevestre
3.1 Introduction . . ............................................ 49
3.2 The One-Way Error Components Model . . . ................... 50
3.2.1 De?nition/Assumptions of the Model . . ............... 50
3.2.2 TheGLSEstimator ................................ 52
3.2.3 The Feasible GLS Estimator . . . . . ................... 55
3.2.4 SomeOtherEstimators............................. 58
3.2.5 Prediction........................................ 63
3.3 More General Structures of the Disturbances . . . ............... 64
3.3.1 The Two-Way Error Components Model . . . . . . ........ 64
3.3.2 Serial Correlation in the Disturbances . . ............... 70
3.3.3 Two-Way Error Components vs Kmenta’s Approach .... 73
3.3.4 Heteroskedasticity in the Disturbances . ............... 74
3.4 Testing .................................................. 78
3.4.1 Testing for the Absence of Individual Effects . . ........ 79
3.4.2 Testing for Uncorrelated Effects: Hausman’s Test . . . .... 80
3.4.3 TestingforSerialCorrelation........................ 81
3.4.4 Testing for Heteroskedasticity . . . . ................... 82
3.5 Estimation Using Unbalanced Panels . . . . . . ................... 84
References . . . . . ................................................ 85
4 Endogenous Regressors and Correlated Effects .................. 89
Rachid Boumahdi and Alban Thomas
4.1 Introduction . . ............................................ 89
4.2 Estimation of Transformed Linear Panel Data Models . . . ........ 90
4.2.1 Error Structures and Filtering Procedures . . . . . . ........ 91
4.2.2 An IV Representation of the Transformed Linear Model . 93
4.3 EstimationwithTime-InvariantRegressors.................... 95
4.3.1 Introduction . ..................................... 95
4.3.2 InstrumentalVariableEstimation..................... 96
4.3.3 More Ef?cient IV Procedures . . . . ................... 98
4.4 A Measure of Instrument Relevance . . . . . . . ................... 99
4.5 Incorporating Time-Varying Regressors .......................101
4.5.1 InstrumentalVariablesEstimation....................102
4.6 GMM Estimation of Static Panel Data Models . . ...............104
4.6.1 Static Model Estimation . . ..........................105
4.6.2 GMM Estimation with HT, AM and BMS Instruments . . 107
4.7 Unbalanced Panels . . . .....................................108
References . . . . . ................................................110
5 The Chamberlain Approach to Panel Data: An Overview and Some
Simulations ................................................ 113
Bruno Cr′ epon and Jacques Mairesse
5.1 Introduction . . ............................................113
5.2 The Chamberlain Π MatrixFramework.......................115Contents ix
5.2.1 The Π Matrix.....................................115
5.2.2 Relations Between Π andtheParametersofInterest.....118
5.2.3 FourImportantCases ..............................120
5.2.4 Restrictions on the Covariance Matrix of the Disturbances124
5.2.5 A Generalization of the Chamberlain Method . . ........125
5.2.6 The Vector Representation of the Chamberlain
EstimatingEquations ..............................126
5.2.7 The Estimation of Matrix Π ........................127
5.3 Asymptotic Least Squares . .................................130
5.3.1 ALSEstimation...................................130
5.3.2 TheOptimalALSEstimator ........................132
5.3.3 Speci?cation Testing in the ALS Framework . . . ........135
5.4 The Equivalence of the GMM and the Chamberlain Methods .....137
5.4.1 A Reminder on the GMM...........................137
5.4.2 Equivalence of the GMM and the Chamberlain Methods . 139
5.4.3 Equivalence in Speci?c Cases . . . . ...................140
5.5 MonteCarloSimulations...................................144
5.5.1 Design of the Simulation Experiments . ...............144
5.5.2 ConsistencyandBias ..............................147
5.5.3 Ef?ciencyandRobustness ..........................152
5.5.4 Standard Errors . . .................................155
5.5.5 Speci?cation Tests . . . . . . . ..........................158
5.6 Appendix A: An Extended View of the Chamberlain Method . ....160
5.6.1 Simultaneous Equations Models . . ...................160
5.6.2 VAR Models .....................................160
5.6.3 Endogenous Attrition . . . . ..........................162
5.7 Appendix B: Vector Representation of the Chamberlain
EstimatingEquations ......................................163
5.7.1 TheVecOperator .................................163
5.7.2 CorrelatedEffects .................................164
5.7.3 ErrorsinVariables.................................164
5.7.4 WeakSimultaneity ................................166
5.7.5 CombinationoftheDifferentCases ..................166
5.7.6 Lagged Dependent Variable . . . . . . ...................167
5.7.7 Restrictions on the Covariance Matrix of the Disturbances167
5.8 Appendix C: Manipulation of Equations and Parameters
intheALSFramework.....................................168
5.8.1 TransformationoftheEstimatingEquations ...........168
5.8.2 Eliminating Parameters of Secondary Interest . . ........169
5.8.3 Recovering Parameters of Secondary Interest
OnceEliminated ..................................170
5.8.4 Elimination of Auxiliary Parameters . . . ...............173
5.9 Appendix D: Equivalence Between Chamberlain’s, GMM
and Usual Panel Data Estimators . . ..........................174
5.10 Appendix E: Design of Simulation Experiments . ...............177x Contents
5.10.1 Generating Process of the Variable x ..................177
5.10.2 Regression Model .................................178
5.10.3 CalibrationofSimulations ..........................179
5.10.4 Three Scenarios . . .................................180
5.10.5 TheChamberlainandGMMEstimators...............180
5.10.6 Standard Errors and Speci?cation Tests ...............181
References . . . . . ................................................181
6 Random Coef?cient Models .................................. 185
Cheng Hsiao and M. Hashem Pesaran
6.1 Introduction . . ............................................185
6.2 The Models . . ............................................186
6.3 Sampling Approach . . .....................................189
6.4 MeanGroupEstimation....................................192
6.5 Bayesian Approach . . . .....................................193
6.6 Dynamic Random Coef?cients Models . . . . ...................197
6.7 Testing for Heterogeneity Under Weak Exogeneity . . . . . ........199
6.8 A Random Coef?cient Simultaneous Equation System . . ........203
6.9 Random Coef?cient Models with Cross-Section
Dependence . . ............................................206
6.10 Concluding Remarks . .....................................208
References . . . . . ................................................211
7 Parametric Binary Choice Models ............................. 215
Michael Lechner, St′ efan Lollivier and Thierry Magnac
7.1 Introduction . . ............................................215
7.2 Random Effects Models Under Strict Exogeneity . . . . . . . ........217
7.2.1 Errors are Independent Over Time ...................218
7.2.2 OneFactorErrorTerms ............................219
7.2.3 General Error Structures . . ..........................221
7.2.4 Simulation Methods . . . . . ..........................223
7.2.5 How to Choose a Random Effects Estimator
foranApplication.................................228
7.2.6 CorrelatedEffects .................................229
7.3 Fixed Effects Models Under Strict Exogeneity . . ...............230
7.3.1 The Model . . .....................................231
7.3.2 The Method of Conditional Likelihood ...............232
7.3.3 Fixed Effects Maximum Score . . . . ...................235
7.3.4 GMMEstimation .................................236
7.3.5 Large-TApproximations ...........................237
7.4 Dynamic Models . . . . . .....................................238
7.4.1 Dynamic Random Effects Models ....................238
7.4.2 Dynamic Fixed Effects Models . . . ...................241
References . . . . . ................................................242Contents xi
Part II Advanced Topics
8 Dynamic Models for Short Panels ............................. 249
Mark N. Harris, L′ aszl ′ oM′ aty′ as and Patrick Sevestre
8.1 Introduction . . ............................................249
8.2 The Model . . . ............................................250
8.3 The Inconsistency of Traditional Estimators ...................252
8.4 IVandGMMEstimators ...................................255
8.4.1 Uncorrelated Individual Effects: The Original
Balestra–NerloveEstimatoranditsExtensions .........256
8.4.2 Correlated Individual Effects . . . . . ...................257
8.4.3 Some Monte Carlo Evidence . . . . . ...................269
8.5 The Maximum Likelihood Estimator . . . . . . ...................270
8.6 Testing in Dynamic Models . . . . . . . ..........................272
8.6.1 TestingtheValidityofInstruments ...................272
8.6.2 Testing for Unobserved Effects . . . ...................273
8.6.3 Testing for the Absence of Serial Correlation in ε .......274
8.6.4 Signi?cance Testing in Two-Step Variants . . . . . ........275
References . . . . . ................................................276
9 Unit Roots and Cointegration in Panels ........................ 279
J¨ org Breitung and M. Hashem Pesaran
9.1 Introduction . . ............................................279
9.2 First Generation Panel Unit Root Tests . . . . ...................281
9.2.1 The Basic Model . .................................281
9.2.2 DerivationoftheTests .............................282
9.2.3 NullDistributionoftheTests........................284
9.2.4 AsymptoticPoweroftheTests ......................287
9.2.5 Heterogeneous Trends . . . ..........................288
9.2.6 Short-Run Dynamics . . . . . ..........................291
9.2.7 Other Approaches to Panel Unit Root Testing . . ........293
9.3 Second Generation Panel Unit Root Tests . . ...................295
9.3.1 Cross-Section Dependence . . . . . . . ...................295
9.3.2 TestsBasedonGLSRegressions.....................296
9.3.3 TestStatisticsBasedonOLSRegressions .............297
9.3.4 Other Approaches .................................298
9.4 Cross-UnitCointegration...................................299
9.5 Finite Sample Properties of Panel Unit Root Tests . . . . . . ........301
9.6 Panel Cointegration: General Considerations . . . ...............302
9.7 Residual-Based Approaches to Panel Cointegration . . . . . ........306
9.7.1 Spurious Regression . . . . . ..........................306
9.7.2 Tests of Panel Cointegration . . . . . . ...................307
9.8 Tests for Multiple Cointegration . . . ..........................308
9.9 Estimation of Cointegrating Relations in Panels . ...............309
9.9.1 SingleEquationEstimators .........................309
9.9.2 SystemEstimators.................................312xii Contents
9.10 Cross-Section Dependence and the Global VAR ................313
9.11 Concluding Remarks . .....................................316
References . . . . . ................................................316
10 Measurement Errors and Simultaneity ......................... 323
Erik Bi?rn and Jayalakshmi Krishnakumar
10.1 Introduction . . ............................................323
10.2 Measurement Errors and Panel Data . . . . . . . ...................323
10.2.1 Model and Orthogonality Conditions . . ...............325
10.2.2 Identi?cation and the Structure of the Second
OrderMoments ...................................327
10.2.3 MomentConditions ...............................328
10.2.4 Estimators Constructed from Period Means . . . . ........331
10.2.5 GMM Estimation and Testing in the General Case . . ....332
10.2.6 Estimation by GMM, Combining Differences and Levels 335
10.2.7 Extensions:Modi?cations ..........................343
10.2.8 Concluding Remarks . . . . . ..........................343
10.3 Simultaneity and Panel Data . . . . . . ..........................344
10.3.1 SEMwithEC.....................................345
10.3.2 Extensions .......................................361
10.4 Conclusion . . . ............................................364
References . . . . . ................................................365
11 Pseudo-Panels and Repeated Cross-Sections .................... 369
Marno Verbeek
11.1 Introduction . . ............................................369
11.2 Estimation of a Linear Fixed Effects Model ...................370
11.3 Estimation of a Linear Dynamic Model . . . . ...................376
11.4 Estimation of a Binary Choice Model . . . . . ...................380
11.5 Concluding Remarks . .....................................381
References . . . . . ................................................382
12 Attrition, Selection Bias and Censored Regressions ............... 385
Bo Honor′ e, Francis Vella and Marno Verbeek
12.1 Introduction . . ............................................385
12.2 Censoring, Sample Selection and Attrition . ...................386
12.3 Sample Selection and Attrition . . . . ..........................389
12.4 Sample Selection Bias and Robustness of Standard Estimators ....391
12.5 Tobit and Censored Regression Models . . . . ...................393
12.5.1 Random Effects Tobit . . . . ..........................394
12.5.2 Random Effects Tobit with Endogenous Explanatory
Variables.........................................396
12.5.3 Dynamic Random Effects Tobit . . . ...................398
12.5.4 FixedEffectsTobitEstimation.......................399
12.5.5 Semi-parametricEstimation.........................401Contents xiii
12.5.6 Semi-parametric Estimation in the Presence
of Lagged Dependent Variables . . . ...................402
12.6 Models of Sample Selection and Attrition . . ...................402
12.6.1 Maximum Likelihood Estimators . ...................403
12.6.2 Two-StepEstimators...............................404
12.6.3 AlternativeSelectionRules .........................407
12.6.4 Two-StepEstimatorswithFixedEffects...............408
12.6.5 Semi-parametric Sample Selection Models . . . . ........409
12.6.6 Semi-parametric Estimation of a Type-3 Tobit Model . . . 410
12.7 SomeEmpiricalApplications ...............................412
12.7.1 Attrition in Experimental Data .......................412
12.7.2 Real Wages Over the Business Cycle . . ...............413
12.7.3 Unions and Wages . . . . . . . ..........................415
References . . . . . ................................................416
13 Simulation Techniques for Panels: Ef?cient Importance Sampling .. 419
Roman Liesenfeld and Jean-Franc ?ois Richard
13.1 Introduction . . ............................................419
13.2 Pseudorandom Number Generation ..........................420
13.2.1 UnivariateDistributions ............................421
13.2.2 MultivariateDistributions...........................424
13.3 Importance Sampling . .....................................426
13.3.1 General Principle . .................................426
13.3.2 Ef?cient Importance Sampling . . . ...................428
13.3.3 MC Sampling Variance of (E)IS Estimates . . . . ........431
13.3.4 GHKSimulator ...................................432
13.3.5 Common Random Numbers . . . . . . ...................432
13.4 Simulation-Based Inference Procedures . . . . ...................434
13.4.1 Integration in Panel Data Models . ...................434
13.4.2 Simulated Likelihood . . . . ..........................435
13.4.3 SimulatedMethodofMoments ......................435
13.4.4 Bayesian Posterior Moments . . . . . ...................437
13.5 Numerical Properties of Simulated Estimators . . ...............437
13.6 EIS Application: Logit Panel with Unobserved Heterogeneity ....439
13.6.1 The Model . . .....................................439
13.6.2 EIS Evaluation of the Likelihood . ...................440
13.6.3 EmpiricalApplication..............................443
13.7 Conclusion . . . ............................................445
13.8 Appendix: Implementation of EIS for the Logit Panel Model . ....446
References . . . . . ................................................448
14 Semi-parametric and Non-parametric Methods in Panel Data
Models .................................................... 451
Chunrong Ai and Qi Li
14.1 Introduction . . ............................................451
14.2 Linear Panel Data Model . . .................................452xiv Contents
14.2.1 AdditiveEffect....................................452
14.2.2 Multiplicative Effect . . . . . ..........................460
14.3 Nonlinear Panel Data Model . . . . . . ..........................462
14.3.1 Censored Regression Model . . . . . . ...................462
14.3.2 Discrete Choice Model . . . ..........................470
14.3.3 Sample Selection Model . . ..........................474
14.4 Conclusion . . . ............................................475
References . . . . . ................................................476
15 Panel Data Modeling and Inference: A Bayesian Primer .......... 479
Siddhartha Chib
15.1 Introduction . . ............................................479
15.1.1 HierarchicalPriorModeling.........................480
15.1.2 ElementsofMarkovChainMonteCarlo ..............483
15.1.3 Some Basic Bayesian Updates . . . . ...................486
15.1.4 Basic Variate Generators . ..........................488
15.2 Continuous Responses .....................................489
15.2.1 Gaussian–Gaussian Model ..........................490
15.2.2 Robust Modeling of bi: Student–Student
and Student-Mixture Models . . . . . ...................492
15.2.3 Heteroskedasticity . . . . . . . ..........................495
15.2.4 SerialCorrelation .................................496
15.3 Binary Responses . . . . .....................................497
15.4 Other Outcome Types . .....................................501
15.4.1 CensoredOutcomes ...............................501
15.4.2 Count Responses . .................................502
15.4.3 Multinomial Responses . . ..........................503
15.5 Binary Endogenous Regressor . . . . ..........................504
15.6 Informative Missingness . . .................................507
15.7 Prediction................................................508
15.8 Residual Analysis . . . . .....................................509
15.9 Model Comparisons . . .....................................509
15.9.1 Gaussian–Gaussian Model ..........................512
15.9.2 Gaussian–Gaussian Tobit model . . ...................512
15.9.3 Panel Poisson Model . . . . . ..........................513
15.10 Conclusion . . . ............................................513
References . . . . . ................................................514
16 To Pool or Not to Pool? ...................................... 517
Badi H. Baltagi, Georges Bresson and Alain Pirotte
16.1 Introduction . . ............................................517
16.2 Tests for Poolability, Pretesting and Stein-Rule Methods .........521
16.2.1 Tests for Poolability . . . . . ..........................521
16.2.2 Pretesting and Stein-Rule Methods ...................525
16.2.3 Example .........................................526
16.3 Heterogeneous Estimators . .................................527Contents xv
16.3.1 AveragingEstimators ..............................529
16.3.2 Bayesian Framework. . . . . ..........................530
16.3.3 AnExample......................................538
16.4 Comments on the Predictive Approach . . . . ...................541
16.4.1 From the Post-sample Predictive Density... ............541
16.4.2 ... to the Good Forecast Performance of the
Hierarchical Bayes Estimator: An Example . . . . ........542
16.5 Conclusion . . . ............................................544
References . . . . . ................................................545
17 Duration Models and Point Processes .......................... 547
Jean-Pierre Florens, Denis Foug` ere and Michel Mouchart
17.1 Marginal Duration Models ..................................548
17.1.1 Distribution, Survivor and Density Functions . . ........548
17.1.2 Truncated Distributions and Hazard Functions . ........550
17.2 Conditional Models . . .....................................552
17.2.1 General Considerations . . . ..........................552
17.2.2 The Proportional Hazard or Cox Model ...............555
17.2.3 The Accelerated Time Model . . . . . ...................557
17.2.4 Aggregation and Heterogeneity . . . ...................558
17.2.5 Endogeneity . .....................................560
17.3 Competing Risks and Multivariate Duration Models . . . . ........561
17.3.1 MultivariateDurations .............................561
17.3.2 Competing Risks Models: De?nitions . ...............563
17.3.3 Identi?ability of Competing Risks Models . . . . . ........566
17.3.4 Right-Censoring ..................................568
17.4 Inference in Duration Models . . . . . ..........................570
17.4.1 Introduction . .....................................570
17.4.2 Parametric Models . . . . . . ..........................570
17.4.3 Non-parametric and Semi-parametric Models . . ........576
17.5 CountingProcessesandPointProcesses ......................579
17.5.1 De?nitions .......................................579
17.5.2 Stochastic Intensity, Compensator and Likelihood
of a Counting Process . . . . ..........................581
17.6 Poisson,MarkovandSemi-MarkovProcesses .................584
17.6.1 PoissonProcesses .................................584
17.6.2 MarkovProcesses .................................585
17.6.3 Semi-MarkovProcesses ............................592
17.7 StatisticalAnalysisofCountingProcesses ....................594
17.7.1 The Cox Likelihood . . . . . ..........................596
17.7.2 The Martingale Estimation of the Integrated Baseline
Intensity .........................................597
17.8 Conclusions . . ............................................600
References . . . . . ................................................600xvi Contents
18 GMM for Panel Data Count Models ........................... 603
Frank Windmeijer
18.1 Introduction . . ............................................603
18.2 GMMinCross-Sections....................................604
18.3 Panel Data Models . . . .....................................606
18.3.1 Strictly Exogenous Regressors .......................607
18.3.2 Predetermined Regressors ..........................608
18.3.3 Endogenous Regressors . . ..........................609
18.3.4 Dynamic Models . .................................610
18.4 GMM...................................................612
18.5 ApplicationsandSoftware..................................614
18.6 Finite Sample Inference . . . .................................615
18.6.1 Wald Test and Finite Sample Variance Correction . . . ....615
18.6.2 Criterion-BasedTests ..............................617
18.6.3 Continuous Updating Estimator . . . ...................618
18.6.4 MonteCarloResults ...............................619
References . . . . . ................................................623
19 Spatial Panel Econometrics ................................... 625
Luc Anselin, Julie Le Gallo and Hubert Jayet
19.1 Introduction . . ............................................625
19.2 SpatialEffects............................................626
19.2.1 SpatialWeightsandSpatialLagOperator .............628
19.2.2 Spatial Lag Model . . . . . . . ..........................630
19.2.3 Spatial Error Model . . . . . . ..........................632
19.3 A Taxonomy of Spatial Panel Model Speci?cations . . . . . ........636
19.3.1 Temporal Heterogeneity . . ..........................637
19.3.2 Spatial Heterogeneity . . . . ..........................639
19.3.3 Spatio-Temporal Models . ..........................644
19.4 Estimation of Spatial Panel Models ..........................648
19.4.1 Maximum Likelihood Estimation . ...................648
19.4.2 InstrumentalVariablesandGMM....................652
19.5 Testing for Spatial Dependence . . . . ..........................654
19.5.1 Lagrange Multiplier Tests for Spatial Lag and Spatial
Error Dependence in Pooled Models . . . ...............655
19.5.2 Testing for Spatial Error Correlation in Panel
Data Models . .....................................655
19.6 Conclusions . . ............................................656
References . . . . . ................................................657
Part III Applications
20 Foreign Direct Investment: Lessons from Panel Data ............. 663
Pierre Blanchard, Carl Gaign′ e and Claude Mathieu
20.1 Introduction . . ............................................663
20.2 A Simple Model of FDI . . . .................................664
20.2.1 AssumptionsandPreliminaryResults.................665Contents xvii
20.2.2 Technology and Country Characteristics
as Determinants of FDI . . . ..........................666
20.3 Econometric Implementation and Data . . . . ...................668
20.3.1 A General Econometric Model . . . ...................669
20.3.2 FDIandDataIssues ...............................670
20.4 EmpiricalEstimations:SelectedApplications..................672
20.4.1 Testing the Trade-Off Between FDI and Exports . . . . ....672
20.4.2 TestingtheRoleofTradePolicyinFDI ...............677
20.4.3 Testing the Relationship Between FDI
and Exchange Rate . . . . . . ..........................683
20.5 Some Recent Econometric Issues . . ..........................690
20.5.1 FDI, Panel Data and Spatial Econometrics . . . . . ........690
20.5.2 Exchange Rate, Unit Roots and Cointegration . . ........691
References . . . . . ................................................693
21 Stochastic Frontier Analysis and Ef?ciency Estimation ........... 697
Christopher Cornwell and Peter Schmidt
21.1 MeasurementofFirmEf?ciency.............................698
21.2 Introduction to SFA . . .....................................700
21.2.1 TheBasicSFAEmpiricalFramework.................700
21.2.2 Stochastic vs Deterministic Frontiers . . ...............700
21.2.3 Other Frontier Functions . ..........................702
21.2.4 SFAwithCross-SectionData........................703
21.3 SFA with Panel Data . .....................................704
21.3.1 Models with Time-Invariant Inef?ciency . . . . . . ........704
21.3.2 Models with Time-Varying Inef?ciency ...............714
21.4 Applications .............................................718
21.4.1 Egyptian Tile Manufacturers . . . . . ...................718
21.4.2 Indonesian Rice Farmers . ..........................720
21.5 Concluding Remarks . .....................................723
References . . . . . ................................................723
22 Econometric Analyses of Linked Employer–Employee Data ....... 727
John M. Abowd, Francis Kramarz and Simon Woodcock
22.1 Introduction . . ............................................727
22.2 APrototypicalLongitudinalLinkedDataSet ..................729
22.2.1 MissingData .....................................730
22.2.2 SamplingfromLinkedData.........................732
22.3 Linear Statistical Models with Person and Firm Effects . . ........733
22.3.1 A General Speci?cation . . ..........................733
22.3.2 The Pure Person and Firm Effects Speci?cation ........734
22.4 De?nition of Effects of Interest . . . . ..........................735
22.4.1 Person Effects and Unobservable Personal
Heterogeneity. . . . .................................735
22.4.2 Firm Effects and Unobservable Firm Heterogeneity . ....736
22.4.3 Firm-AveragePersonEffect.........................737xviii Contents
22.4.4 Person-AverageFirmEffect.........................737
22.4.5 Industry Effects . . .................................738
22.4.6 OtherFirmCharacteristicEffects ....................739
22.4.7 Occupation Effects and Other Person × × × Firm
Interactions.......................................739
22.5 Estimation by Fixed Effects Methods . . . . . . ...................739
22.5.1 Estimation of the Fixed Effects Model by Direct
Least Squares . . . .................................739
22.5.2 Consistent Methods for β and γ (The Firm-Speci?c
Returns to Seniority) . . . . . ..........................743
22.6 The Mixed Model . . . . .....................................744
22.6.1 REML Estimation of the Mixed Model ...............746
22.6.2 Estimating the Fixed Effects and Realized
Random Effects . . .................................747
22.6.3 Mixed Models and Correlated Random Effects Models . . 748
22.7 Models of Heterogeneity Biases in Incomplete Models . ........750
22.7.1 OmissionoftheFirmEffects........................750
22.7.2 OmissionofthePersonEffects ......................751
22.7.3 Inter-industry Wage Differentials . ...................752
22.8 Endogenous Mobility ......................................753
22.8.1 A Generalized Linear Mixed Model . . . ...............754
22.8.2 A Model of Wages, Endogenous Mobility and
ParticipationwithPersonandFirmEffects ............755
22.8.3 Stochastic Assumptions . . ..........................756
22.9 Conclusion . . . ............................................758
References . . . . . ................................................758
23 Life Cycle Labor Supply and Panel Data: A Survey .............. 761
Bertrand Koebel, Franc ?ois Laisney, Winfried Pohlmeier and Matthias
Staat
23.1 Introduction . . ............................................761
23.2 The Basic Model of Life Cycle Labor Supply . . . ...............762
23.2.1 TheFramework...................................763
23.2.2 First Speci?cations of the Utility Function . . . . . ........765
23.3 Taking Account of Uncertainty and Risk . . . ...................768
23.3.1 FirstDevelopments ................................768
23.3.2 Recent Contributions . . . . ..........................770
23.3.3 EmpiricalResults .................................773
23.4 VoluntaryandInvoluntaryNon-participation ..................774
23.4.1 Accounting for the Participation Decision . . . . . ........775
23.4.2 Unemployment ...................................778
23.5 AlternativeParameterizationandImplications .................779
23.6 Relaxing Separability Assumptions ..........................783
23.6.1 Relaxing Within-Period Additive Separability . . ........783
23.6.2 Relaxing Intertemporal Separability in Preferences . ....784Contents xix
23.7 Conclusion . . . ............................................790
References . . . . . ................................................791
24 Dynamic Policy Analysis ..................................... 795
Jaap H. Abbring and James J. Heckman
24.1 Introduction . . ............................................795
24.2 PolicyEvaluationandTreatmentEffects ......................796
24.2.1 TheEvaluationProblem............................796
24.2.2 The Treatment Effect Approach. . . ...................800
24.2.3 Dynamic Policy Evaluation . . . . . . ...................801
24.3 Dynamic Treatment Effects and Sequential Randomization . . ....803
24.3.1 Dynamic Treatment Effects . . . . . . ...................803
24.3.2 Policy Evaluation and Dynamic Discrete-Choice
Analysis .........................................810
24.3.3 TheInformationStructureofPolicies.................813
24.3.4 Selection on Unobservables . . . . . . ...................815
24.4 The Event-History Approach to Policy Analysis . ...............816
24.4.1 Treatment Effects in Duration Models . ...............817
24.4.2 Treatment Effects in More General Event-History
Models . . . . . .....................................823
24.4.3 A Structural Perspective . . ..........................828
24.5 Dynamic Discrete Choice and Dynamic Treatment Effects . . . ....829
24.5.1 Semi-parametric Duration Models and Counterfactuals . . 831
24.5.2 A Sequential Structural Model with Option Values . . ....844
24.5.3 Identi?cationatIn?nity ............................850
24.5.4 Comparing Reduced-Form and Structural Models . . ....851
24.5.5 A Short Survey of Dynamic Discrete-Choice Models . . . . 853
24.6 Conclusion . . . ............................................857
References . . . . . ................................................857
25 Econometrics of Individual Labor Market Transitions ............ 865
Denis Foug` ere and Thierry Kamionka
25.1 Introduction . . ............................................865
25.2 Multi-spell Multi-state Models . . . . ..........................867
25.2.1 General framework . . . . . . ..........................867
25.2.2 Non-parametricandParametricEstimation ............872
25.2.3 Unobserved Heterogeneity . . . . . . . ...................878
25.3 MarkovProcessesUsingDiscrete-TimeObservations...........882
25.3.1 The Time-Homogeneous Markovian Model ............883
25.3.2 The Mover-Stayer Model . ..........................893
25.4 Concluding Remarks . .....................................901
References . . . . . ................................................902xx Contents
26 Software Review ............................................ 907
Pierre Blanchard
26.1 Introduction . . ............................................907
26.2 General-Purpose Econometric Packages . . . ...................908
26.2.1 EViews(v.5.1) ...................................908
26.2.2 LIMDEP(v.8)withNLOGIT(v.3) ..................912
26.2.3 RATS(v.6) ......................................916
26.2.4 SAS(v.9.1) ......................................920
26.2.5 Stata(v.9) .......................................923
26.2.6 TSP(v.5)........................................927
26.3 High-Level Matrix Programming Languages . . . ...............930
26.3.1 GAUSS(v.5).....................................930
26.3.2 Ox(v.3.4) .......................................936
26.4 Performance Hints and Numerical Accuracy Evaluation . ........941
26.4.1 Speed Comparison . . . . . . ..........................941
26.4.2 Numerical Accuracy Evaluations . ...................944
References . . . . . ................................................949