Introduction to Econometrics THIRD EDITION James H. Stock Harvard University Mark W. Watson Princeton University Addison-Wesley Boston Columbus Indianapolis New York San Francisco Upper Saddle River Amsterdam Cape Town Dubai London Madrid Milan Munich Paris Montrea) Toronto Dethi Mexico City Sao Paulo Sidney Hong Kong Seoul Singapore TaipeiEditoria} Director: Sally Yagen Am Director: Anthony Gemmellaro Editor-in-Chief: Donna Battista Cover Designer: Ambony Gemmellara Senior Acquisitions Editor: Adrienne D'Ambrosio. Full-Service Project Management, Art Studio, and Assistant Editor: Jill Kolongowski Electronic Composition: Nesbitt Graphics, In. Director of Marketing: Patrice Jones Printer/Binder: Edwards Brothers Managing Editor: Nancy Fenton Cover Printer: Lehigh-Phoenix Color / Hagerstown Production Coordinator: Alison Eusden_ ‘Text Font: 10/14 Times Ten Roman Senior Manufacturing Buyer: Carol Melville Manager, Rights and Permissions: Michael Joyce About the cover: The cover shows a heat chart of 270 monthly vatiables measuring different aspects of employment, production, income, and sales for the United States, 1974-2010. Each horizontal ine depicts a different variable, and the horizontal axis is the date. Strong monthly increases in a variable are blue and sharp monthly declines are red. The simultaneous dectines in many of these measures during recessions appeat in the figure as vertical red bands. Credits and acknowledgments borrowed from other sources and reproduced, with permission, in this textbook appear on appropriate page within text. Copyright © 2011, 2007,2003 Pearson Education, Inc., publishing as Addison-Wesley. Al rights reserved, Manufactured in the United States of America. This publication is protected by Copyright, and permission should be obtained from the publisher prior 1o any prohibited reptaduction, storage in a retrieval system. or transmission in any form or by any means, electronic, mechanical, photocopying, recording, or likewise. To obtain permission(s) to use material from this work, please submit a written request to Pearson Education, Inc., Rights and Contracts Department, 501 Boylston Street, Suite 900, Boston, MA 92116, fax your request to 617-671-3447, or e-mail at http:!/www-pearsoned.com/legal/permissions htm. Library of Congress Cataloging-in-Publication Data Stock, James Hi Introduction zo econometrics/lames H. Stock, Mark W. Watson, ~ 3rd ed. p.cm. — (The Addison-Wesley series in economics) Includes bibliographical references and index. ISBN-13: 978-0-13-8100900-7 (alk. paper) ISBN-10: 0-13-800900-7 (alk. paper) 1. Econometrics I. Watson, Mark W. I. ttl ¥HBI39,S765 2000 330.01°5195—de22 1098765432 1—BB~14 13 121110 Addison-Westey is an imprint of Sey ISBN-13:978- 0-13-800900-7 www:pearsonhighered.com ISBN-10: 0-13-800900.7Brief Contents PART ONE introduction and Review CHAPTER 1 Economic Questions and Data 1 CHAPTER 2 Review of Probability 14 CHAPTER 3 Review of Statistics 64 PART TWO. Fundamentals of Regression Analysis CHAPTER 4 Linear Regression with One Regressor 107 CHAPTER 5 Regression with a Single Regressor: Hypothesis Tests and Confidence Intervals 144 CHAPTER 6 —_Linear Regression with Multiple Regressors 179 CHAPTER 7 Hypothesis Tests and Confidence Intervals in Multiple Regression 214 CHAPTER 8 Nonlinear Regression Functions 252 CHAPTER9 —_—_ Assessing Studies Based on Multiple Regression 312 PARTTHREE Further Topics in Regression Analysis CHAPTER 10 Regression with Panel Data 347 CHAPTER 11 Regression with a Binary Dependent Variable 381 CHAPTER 12 Instrumental Variables Regression 419 CHAPTER 13 Experiments and Quasi-Experiments 469 PART FOUR Regression Analysis of Economic Time Series Data CHAPTER 14 Introduction to Time Series Regression and Forecasting 516 CHAPTER 15 _ Estimation of Dynamic Causal Effects 583 CHAPTER 16 Additional Topics in Time Series Regression 631 PART FIVE The Econometric Theory of Regression Analysis CHAPTER 17 The Theory of Linear Regression with One Regressor 669 CHAPTER 18 The Theary of Multiple Regression 697Contents Preface xxix PART ONE CHAPTER 1 of 12 13 CHAPTER 2 21 2.2 Introduction and Review Economic Questions and Data 1 Economic Questions We Examine 1 Question #1: Does Reducing Class Size Improve Elementary School Education? 2 Question #2: Is There Racial Discrimination in the Market for Home Loans? 3 Question #3: How Much Do Cigarette Taxes Reduce Smoking? 3 Question #4: What Will the Rate of inflation Be Next Year? 4 Quantitative Questions, Quantitative Answers 5 Causal Effects and Idealized Experiments 5 Estimation of Causal Effects 6 Forecasting and Causality 7 Data: Sources and Types 7 Experimental Versus Observational Data 7 Cross-Sectional Data 8 Time Series Data 9 Panel Data 11 Review of Probability 14 Random Variables and Probability Distributions 15 Probabilities, the Sample Space, and Random Variables 15 Probability Distribution of a Discrete Random Variable 16 Probability Distribution of a Continuous Random Variable 18 Expected Values, Mean, and Variance 18 The Expected Value of a Random Variable 18 ‘The Standard Deviation and Variance 21 Mean and Variance of a Linear Function of a Random Variable 22 Other Measures of the Shape of a Distribution 23 viivili Conteris 23 24 25 26 CHAPTER 3 34 3.2 Two Random Variables 26 Joint and Marginal Distributions 26 Conditional Distributions 27 Independence 31 Covariance and Correlation 31 ‘The Mean and Variance of Sums of Random Variables 32 The Normal, Chi-Squared, Student ¢, and F Distributions 36 The Normal Distribution 36 The Chi-Squared Distribution 41 The Student ¢ Distribution 4% The F Distribution 42 Random Sampling and the Distribution of the Sample Average 43 Random Sampling 43 The Sampling Distribution of the Sample Average 44 Large-Sample Approximations to Sampling Distributions 47 ‘The Law of Large Numbers and Consistency 48 ‘The Central Limit Theorem 50 APPENDIX 2.1 Derivation of Results in Key Concept2.3 62 Review of Statistics Estimation of the Population Mean 65 Estimators and Their Properties 65 Properties of Y 67 The Importance of Random Sampling 69 Hypothesis Tests Concerning the Population Mean 70 Null and Alternative Hypotheses 70 The pValue “71 Calculating the pValue When oy ls Known 72 The Sample Variance, Sample Standard Deviation, and Standard Error 73 Calculating the p-Value When oy Is Unknown 75 The tStatisic 75 Hypothesis Testing with a Prespecified Significance Level 76 One-Sided Alternatives 7833 34 3.5 3.6 37 PART TWO. CHAPTER 4 Al 42 43 Contents ix Confidence Intervals for the Population Mean 79 Comparing Means from Different Populations 81 Hypothesis Tests for the Difference Between Two Means 81 Confidence Intervals for the Difference Between Two Population Means 83 Differences-of-Means Estimation cf Causal Effects Using Experimental Data 83 ‘The Causal Effect as a Difference of Conditional Expectations 84 Estimation of the Causal Effect Using Differences of Means 84 Using the ¢-Statistic When the Sample Size Is Small 86 The &-Statistic and the Student ¢ Distribution 86 Use of the Student £ Distribution in Practice 90 Scatterplots, the Sample Covariance, and the Sample Correlation 91 Scatterplots 91 ‘Sample Covariance and Correlation 91 APPENDIX 3.1 The U.S. Current Population Survey 104 APPENDIX 3.2 Two Proofs That ¥ Is the Least Squares Estimator of 104 ‘APPENDIX 3.3 A Proof That the Sample Variance Is Consistent 105 Fundamentals of Regression Analysis Linear Regression with One Regressor 107 The Linear Regression Model 107 Estimating the Coefficients of the Linear Regression Model 112 The Ordinary Least Squares Estimator 114 OLS Estimates of the Relationship Between Test Scores and the Student-Teacher Ratio 115 Why Use the OLS Estimator? 117 Measures of Fit 119 The R219 The Standard Error of the Regression 120 Application to the Test Score Data 121x Contents 44 45 4.6 CHAPTER 5. 5.1 5.5 The Least Squares Assumptions 122 Assumption #1: The Conditional Distribution of u; Given X; Has a Mean of Zero 122 Assumption #2: (X,, ¥),7= Distributed 124 Assumption #3: Large Outliers Are Unlikely 125 ‘Use of the Least Squares Assumptions 126 Sampling Distribution of the OLS Estimators 127 ‘The Sampling Distribution of the OLS Estimators 128, n, Are Independently and Identically Conclusion 131 APPENDIX 4. The California Test Score Data Set 139 APPENDIX 4.2 Derivation of the OLS Estimators 139 APPENDIX 4.3 Sampling Distribution of the OLS Estimator 140 Regression with a Single Regressor: Hypothesis Tests and Confidence Intervals 144 Testing Hypotheses About One of the Regression Coefficients 144 ‘Two-Sided Hypotheses Concerning B, 145 One-Sided Hypotheses Concerning f, 148 Testing Hypotheses About the Intercept By 150 Confidence Intervals for a Regression Coefficient 151 Regression When X Is a Binary Variable 153 Inierpretation of the Regression Coefficients 153 Heteroskedasticity and Homoskedasticity 155 What Are Heteroskedasticity and Homoskedasticty? 156 Mathematica implications of Homoskedasticity 158 What Does This Mean in Practice? 159 The Theoretical Foundations of Ordinary Least Squares 161 Linear Conditionally Unbiased Estimators and the Gauss~Markov Theorem 162 Regression Estimators Other Than OLS 1635.6 5.7 CHAPTER 6 61 6.2 6.4 65 6.6 Contents xi Using the ¢-Statistic in Regression When the Sample Size Is Small 164 The Statistic and the Student ¢ Distribution 164 Use of the Student ¢ Distribution in Practice 165 Conclusion 166 APPENDIX 5.1 Formulas for OLS Standard Errors 174 APPENDIX 5.2 The Gauss-Markov Conditions and a Proof of the Gauss-Markov Theorem 175, Linear Regression with Multiple Regressors 179 Omitted Variable Bias 179 Definition of Omitted Variable Bias 180 A Formula for Omitted Variable Bias 182 Addressing Omitted Variable Bias by Dividing the Data into Groups 184 The Multiple Regression Model 186 The Population Regression Line 186 The Population Multiple Regression Model 187 The OLS Estimator in Multiple Regression 189 The OLS Estimator 190 Application to Test Scores and the Student-Teacher Ratio 191 Measures of Fit in Multiple Regression 193 The Standard Error of the Regression (SER) 193 The R2 193 The "Adjusted R2 194 Application to Test Scores 195 The Least Squares Assumptions in Multiple Regression 196 Assumption #1: The Conditional Distribution of u; Given Xy, Xo... Xj Has a Mean of Zero 196 Assumption #2: Xi, Xop Xue YO/= 1... Areiid. 196 Assumption #3: Large Outliers Are Unlikely 196 Assumption #4: No Perfect Multicollinearity 197 The Distribution of the OLS Estimators in Multiple Regression 198xii Contents 67 68 CHAPTER 7 7 72 73 74 75 76 a7 Mutticollinearity 199 Examples ot Perfect Mutticallinearity 200. Imperfect Multicollinearity 202 Conclusion 203 APPENDIX 6.1 Derivation of Equation (6.1) 211 APPENDIX 6.2. Distribution of the OLS Estimators When There Are Two Regressors and Homoskedastic Errors 212 APPENDIX 6.3 The Frisch-Waugh Theorem 212 Hypothesis Tests and Confidence Intervals in Multiple Regression 214 Hypothesis Tests and Confidence Intervals for a Single Coefficient 214 Standard Errors for the OLS Estimators 214 Hypothesis Tests for a Single Coefficient 215 Confidence Intervais for a Single Coefficient 216 Application to Test Scores and the Student—Teacher Ratio 217 Tests of Joint Hypotheses 219 Testing Hypotheses on Two or More Coefficients 219 The FStatistic 221 ‘Application to Test Scores and the Student~Teacher Ratio 223 The Homoskedasticity-Only F-Statistic 224 Testing Single Restrictions Involving Multiple Coefficients 226 Confidence Sets for Multiple Coefficients 228 Model Specification for Multiple Regression 229 Omitted Variable Bias in Multiple Regression 230 The Role of Controi Variables in Multiple Regression 230 ‘Model Specification in Theory and in Practice 233 Interpreting the R? and the Adjusted R? in Practice 234 Analysis of the Test Score Data Set 235 Conclusion 240CHAPTER 8 81 82 83 B84 85 CHAPTER 9 91 Contents xiii APPENDIX 71 The Bonferroni Test of a Joint Hypothesis. 248 APPENDIX 22 Conditional Mean Independence 250 Nonlinear Regression Functions 252 ‘A General Strategy for Modeling Nonlinear Regression Functions 254 Test Scores and District Income 254 The Effect on Y of a Change in X in Nonlinear Specifications 257 ‘A General Approach to Modeling Noniinearities Using Multiple Regression 262 Nonlinear Functions of a Single Independent Variable 262 Polynomials 263 Logarithms 265 Polynomial and Logarithmic Models of Test Scores and District Income 273 Interactions Between Independent Variables 274 Interactions Between Two Binary Variables 275 Interactions Between a Continuous and a Binary Variable 278 Interactions Between Two Continuous Variables 282 Nonlinear Effects on Test Scores of the Student-Teacher Ratio 286 Discussion of Regression Results 289 Summary of Findings 293 Conclusion 294 APPENDIX 8.1 Regression Functions That Are Nonlinear in the Parameters 306 APPENDIX 8.2 Slopes and Elasticities for Nonlinear Regression Functions 309 Assessing Studies Based on Multiple Regression 312 Internal and Extemal Validity 312 Threats to Internal Validity 313, Threats to External Validity 314xiv Contents 9.2 9.3 94 95 PART THREE CHAPTER 10 10.1 a i) 10.4 Threats to Internal Validity of Multiple Regression Analysis 316 Omitted Variable Bias 316 Misspecification of the Functional Form of the Regression Function 318 Measurement Error and Errors.in-Variables Bias 319 Missing Data and Sample Selection 322 Simultaneous Causality 324 Sources of Inconsistency of OLS Standard Errors 326 Internal and External Validity When the Regression Is Used for Forecasting 327 Using Regression Models for Forecasting 327 Assessing the Validity of Regression Models for Forecasting 329 Example: Test Scores and Class Size 329 External Validity 329 Internal Validity 336 Discussion and implications 338 Conclusion 339 APPENDIX 9.1 The Massachusetts Elementary School Testing Data 345 Further Topics in Regression Analysis Regression with Panel Data 347 Panel Data 348 Example: Traffic Deaths and Alcohol Taxes 349 Panel Data with Two Time Periods: “Before and After” Comparisons 351 Fixed Effects Regression 354 ‘The Fixed Effects Regression Model 354 Estimation and Inference 356 Application to Traffic Deaths 358 Regression with Time Fixed Effects 358 Time Effects Only 359 Both Entity and Time Fixed Effects 360.10.5 10.6 10.7 CHAPTER 11 11 11.2 i143 114 115 CHAPTER 12 12.1 Contents xv The Fixed Effects Regression Assumptions and Standard Errors for Fixed Effects Regression 362 The Fixed Effects Regression Assumptions 362 Standard Errors for Fixed Effects Regression 364 Drunk Driving Laws and Traffic Deaths 365 Conclusion 369 APPENDIX 10.1 The State Traffic Fatality Data Set 376 APPENDIX 10.2 Standard Errors for Fixed Effects Regression 376 Regression with a Binary Dependent Variable 381 Binary Dependent Variables and the Linear Probability Model 382 Binary Dependent Variables 382 ‘The Linear Probability Model 384 Probit and Logit Regression 387 Probit Regression 387 Logit Regression 392 ‘Comparing the Linear Probability, Probit, and Logit Models 394 Estimation and Inference in the Logit and Probit Models 394 Nonlinear Least Squares Estimation 395 ‘Maximum Likelihood Estimation 396 Measures of Fit 397 Application to the Boston HMDA Data 398 Conclusion 405 APPENDIX 11.1 The Boston HMDA Data Set 413 APPENDIX 11.2 Maximum Likelihood Estimation 413 APPENDIX 11.3 Other Limited Dependent Variable Models 416 Instrumental Variables Regression a9 The IV Estimator with a Single Regressor and a Single Instrument 420 The IV Model and Assumptions 420 The Two Stage Least Squares Estimator 421xvi Contents 12.2 12.3 12.4 12.5 CHAPTER 13 BA Why Does IV Regression Work? 422 The Sampling Distribution of the TSLS Estimator 426 Application to the Demand for Cigarettes 428 The General lV Regression Model 430 TSLS in the General IV Model 432 Instrument Relevance and Exogeneity in the General IV Model 433 ‘The IV Regression Assumptions and Sampling Distribution of the TSLS Estimator 434 Inference Using the TSLS Estimator 435 Application to the Demand for Cigarettes 436 Checking Instrument Validity 437 Assumption #1: Instrument Relevance 438 Assumption #2: Instrument Exogeneity 440 Application to the Demand for Cigarettes 443 Where Do Valid Instruments Come From? 448 Three Examples 449 Condusion 453 APPENDIX 12.1 The Cigarette Consumption Panel Data Set 460 APPENDIX 12.2 Derivation of the Formula for the TSLS Estimator in Equation (12.4) 461 APPENDIX 12.3. Large-Sample Distribution of the TSLS éstimator 461 APPENDIX 12.4 Large-Sample Distribution of the TSLS Estimator When the Instrument Is Not Valid 463 APPENDIX 12.5 Instrumental Variables Analysis with Weak Instruments 464 APPENDIX 12.6 TSLS with Control Variables 467 Experiments and Quasi-Experiments 469 Potential Outcomes, Causal Elfects, and idealized Experiments 470 Potential Outcomes and the Average Causal Effect 470 Econometric Methods for Analyzing Experimental Data 47213,2 13.3 13.4 13.5 13.7 PART FOUR CHAPTER 14 141 14.2 Contents xvii Threats to Validity of Experiments 473 Threats to Internal Validity 473 Threats to External Validity 477 Experimental Estimates of the Effect of Class Size Reductions 478 Experimental Design 478 Analysis of the STAR Data 480 ‘Comparison of the Observational and Experimental Estimates of Class Size Effects 485 Quasi-Experiments 487 Examples 488 The Differences-in-Differences Estimator 490 Instrumental Variables Estimators 494 Regression Discontinuity Estimators 494 Potential Problems with Quasi-Experiments 496 Threats to Internal Validity 496 Threats to External Valiaity 498 Experimental and Quasi-Experimental Estimates in Heterogeneous Populations 498 ‘OLS with Heterogeneous Causal Effects 499 IV Regression with Heterogeneous Causal Effects 500 Conclusion 503 APPENDIX 13.1 The Project STAR Data Set S12 APPENDIX 13.2 IV Estimation When the Causal Effect Varies Across Individuals 513 APPENDIX 13.3 The Potential Outcomes Framework for Analyzing Data from Experiments 514 Regression Analysis of Economic Time Series Data Introduction to Time Series Regression and Forecasting 516 Using Regression Models for Forecasting 517 Introduction to Time Series Data and Serial Correlation 518 The Rates of Inflation and Unemployment in the United States 518