Introduction to Econometrics

Transcription

1 Introduction to Econometrics James H. Stock HARVARD UNIVERSITY Mark W. Watson PRINCETON UNIVERSITY... ~ Boston San Francisco New York London Toronco Sydney ToJ..:yo Singapore Madrid Mexico City Munich Paris Cape Town Hong Kong Montreal

2 Brief Contents PART ONE CHAPTE R I CHAPTER 2 CHAPTER 3 Introduction and Review Economic Questions and Data 3 ReviewofProbability 17 Review ofstatistics 65 PART TWO Fundamentals of Regression Analysis 109 CHAPTER 4 Linear Regression with One Regressor 1II CHAPTER 5 Regression with a Single Regressor: Hypothesis Tests and Confidence In tervals 148 CHAPTER 6 Linear Regression with Multiple Regressors 186 CHAPTER 7 CHAPTER 8 CHAPTER 9 Hypothesis Tests and Confidence Intervals in Mult ipl e Regression 220 Nonlinear Regression Functions 254 Assessing Studies Based on Multiple Regression 312 PART THREE Further Topics in Regression Analysis 347 CHAPTER 10 Regression with Panel Data 349 CHAPTER II Regression with a Binary Dependent Variable 383 C HAPTER 12 Instrumental Variables Regression 421 CHAPTER 13 Experiments and Quasi-Experiments 468 PART FOUR CHAPTER 14 Regression Analysis of Economic Time Series Data 523 Introduction to T ime Series Regression and Forecasting 525 CHAPTER 15 Estimation of Dynamic Causal Effects 591 C HAPTER 16 Additional Topics in Time Series Regression 637 PART FIVE The EconometricTheory ofregression Analysis 675 CHAPTER 17 The Theory of Linear Regression wit h One Regressor 677 CHAPTER 18 The T heory of Multiple Regression 704 v

3 Contents Preface XXVIl PART ONE Introduction and Review CHAPTER I Economic Questions and Data Economic Questions We Examine 4 Question # 1: Does Reducing Class Size Improve Elementary School Education? 4 Question # 2: Is There Racial Discrimination in the Market for Home Loans? Question # 3: How Much Do Cigarette Taxes Reduce Smoking? 5 Question #4: W hat Will the Rate of Inflation Be Next Year? 6 Quantitative Questions, Quantitative Answers Causal Effects and Idealized Experime nts 8 Estimation ofcausal Effects 8 Forecasting and Causality Data: Sources and Types 10 Experimental versus Observational Data Cross-Sectional Data [I Time Series Data II Panel Data CHAPTER 2 Review of Probability Random Variables and Probability Distributions 18 Probabilities, the Sample Space, and Random Variables 18 Probability Distribution ofa Discrete Random Variable 19 Probability Distribution ofa Continuous Random Variable Expected Values, Mean, and Variance 23 The Expected Value ofa Random Variable 23 The Standard Deviation and Variance 24 Mean and Variance ofa Linear Function ofa Random Variable Other Measures ofthe Shape ofa Distribution Two Random Variables 29 Joint and Marginal Distributions Conditional Distributions vii

4 viii CO NTENTS 2.4 Independence 34 Covariance and Correlation 34 The Mean and Variance ofsums of Random Variables 35 The Normal, Chi-Squared,Student t, and F Distributions The Normal Distribution 39 The Chi-Squared Distribution 43 The Student t Distribution 44 The F Distribution Random Sampling and the Distribution ofthe Sample Average Random Sampling 45 The Sampling Distribution of the Sample Average Large-Sample Approximations to Sampling Distributions The Law of Large Numbers and Consistency 49 The Central Li mit Theorem 52 APPENDIX 2.1 Derivation of Results in Key Concept CHAPTER Review ofstatistics 65 Estimation ofthe Population Mean 66 Estimators and Their Properties 67 Properties of Y 68 The Importance of Random Sampling 70 Hypothesis Tests Concerning the Population Mean 71 Null and Alternative Hypotheses 72 The p-value 72 Calculating thep-value When O' y Is Known 74 The Sample Variance, Sample Standard Deviation, and Standard Error Calculating thep-value When O'y Is Unknown 76 The t-statistic 77 Hypothesis Testing with a Prespecifled Significance Level 78 One-Sided Alternatives 80 Confidence Interva ls for the Population Mean 81 Comparing Means from Different Populations 83 Hypothesis Tests for the Difference Between Two Means 83 Confidence Intervals for the Difference Between Two Population Means Differences-of-Means Estimat ion ofcausal Effects Using Experimental Data 85 The Causal Effect as a Difference of Conditional Expectations 85 Estimation of the Causal Effec t Using Differences of Means

5 CONTENTS ix 3.6 Using the t-statistic W hen the Sample Size Is Small 88 T he t-statistic and the Student t Distribution 88 Use ofthe Student t Distribution in Practice Scatterplot, the Sample Covariance, and the Sample Correlation 92 Scatterp[ots 93 Sample Covariance and Correlation 94 APPENDIX 3. 1 The U.S. Current Population Survey [05 APPENDIX 3.2 Two Proofs That Y [s the Least Squares Estimator ofj.ly [06 APPENDIX 3.3 A ProofThat the Sample Variance Is Consistent [07 PART TWO Fundamentals of Regression Analysis 109 CHAPTER 4 linear Regression w ith One Regressor III 4. 1 The Linear Regression Model Estimating the Coefficients of the Linear Regression Model 116 T he Ordinary Least Squares Estimator [18 OLS Estimates ofthe Re [ationship Between Test Scores and the Student-Teacher Ratio 120 Why Use the OLS Estimator? Measures of Fit 123 The R2 [23 The Standard Error of the Regression 124 Application to the Test Score Data T he Least Squares Assumptions 126 Assumption #1: The Conditional Distribution ofu, Given X, Has a Mean ofzero [26 Assumption #2: (X,. Y,). i = [..... n Are Independent[y and Ide ntically Distributed [28 Assumption #3: Large Outliers Are Unlike[y [29 Use of the Least Squares Assumptions The Sampling Distribution of the OLS Estimators 131 T he Sampling Distribution ofthe OLS Estimators Conclusion 135 APP ENDIX 4. 1 The California Test Score Data Set [43 A PPEND IX 4. 2 Derivation of the OLS Estimators 143 A PPEN DIX 4. 3 Sampling Di stribution of the O LS Estimator 144

6 x CO NTENTS II CHAPTER 5 Regression with a Single Regressor: Hypothesis Tests and Confidence Intervals Testing Hypotheses About One ofthe Regression Coefficients 149 Two-Sided Hypotheses Concerning f O ne-sided Hypotheses Concerning f Testing Hypotheses About the Intercept f Confidence Intervals for a Regression Coefficient ISS 5.3 Regression W hen X Is a BinaryVariable 158 I. Interpretation ofthe Regression Coefficients 158 I 5.4 Heteroskedasticity and Homoskedasticity 160 W hat Are Heteroskedasticity and Homoskedasticity? 160 Mathematical Implications of Homoskedasticity 163 W hat Does This Mean in Practice? The Theoretical Foundations ofordinary Least Squares 166 Linear Conditionally Unbiased Estimators and the Gauss-Markov -r heorem Regression Estimators Other Than OLS Using the t-statistic in Regression When the Sample Si z~ Is Small 169 The t-statistic and the Student t Distribution 170 Use ofthe Student t Distribution in Practice Conclusion 171 APPE NDI X 5.1 Formulas for OLS Standard Errors 180 APPEND IX 5.2 The Gauss-Markov Conditions and a Proof of the Gauss-Markov Theorem 182 CHAPTER 6 Linear Regression with Multiple Regressors Omitted Variable Bias 186 Definition ofomitted Variable Bias A Formula for Omitted Va riable Bias Addressing Omitted Variable Bias by Dividing the Data into Groub s 6.2 The Multiple Regression Model 193 The Population Regression Line 193 The Population Multiple Regre ssion Model The OLS Estimator in Multiple Regression 196 The OLS Esti mator 197 Appl ication to Test Scores and the Student-Teacher Ratio

7 CONT ENTS xi 6.4 Measures of Fit in Multiple Regression 200 T he Standard Error ofthe Regression (SER) 200 The R2 200 The 'I\djusred R2" 20 I Application to Test Scores The Least Squares Assumptions in Multiple Regression 202 Assumption # 1: The Conditional Distribution ofu,given X I, ' X 2i,...,X k, Has a Mean ofzero 203 Assumption # 2: (X I,' X 2 "...,X iu ' Y,) i =I,...,n Are i.i.d. 203 Assum ption # 3: Large Outliers Are Unlikely 203 Assumption #4: No Perfect Mul ticollinearity The Distribution ofthe OLS Estimators in Multiple Regression Multicollinearity 206 Examples of Perfect Multicollinearity Imperfect Multicollinearity Conclusion APPENDIX 6.1 Derivation of Eq uation (6.1) 218 APPEN DIX 6.2 Distribution of the OLS Estimators W hen There Are Two Regressors and Homoskedastic Errors 218 CHAPTER 7 Hypothesis Tests and Confidence Intervals in Multiple Regression Hypothesis Tests and Confidence Intervals fo r a Single Coefficient 221 Standard Errors for the OLS Estimators 22 1 Hypothesis Tests for a Si ngle Coeffi cient 22 1 Confidence Intervals for a Single Coefficient 223 Application to Test Scores and the Student- Teacher Ratio 7.2 Tests ofjoint Hypotheses 225 Testing Hypotheses on Two or More Coefficients 225 T he F-Statistic 227 Application to Test Scores and the Student- Teacher Ratio T he Homoskedastici ry-only F-Statistic Testing Si ngle Restrictions Involving Multiple Coefficients Confidence Sets for Multiple Coefficients

8 xii CONTE NTS 7.S Model Specification for Multiple Regression 235 Omitted Variable Bias in Multiple Regression 236 Model Specification in Theory and in Practice 236 Interpreting the R2 and the Adjusted R2 in Practice 237 Analysis of the Test Score Data Set 239 Conclusion 244 APPE ND IX 7. 1 The Bonferroni Test ofa Joint Hypotheses 251 CHAPTER CHAPTER Nonlinear Regression Functions 254 A General Strategy for Modeling Nonlinear Regression Functions 256 Test Scores and District Income 256 The Effect on Yofa Change in X in Nonlinear Specifications 260 A General Approach to Modeling Nonlinearities Using Multiple Regression Nonlinear Functions ofa Single Independent Variable 264 Polynomials 265 Logarithms 267 Polynomial and Logarithmic Models oftest Scores and District Income Interactions Between Independent Variables 277 Interactions Between Two Binary Variable s 277 Interactions Between a Continuous and a Binary Variable 280 Interactions Between Two Continuous Variables 286 Nonlinear Effects on Test Scores ofthe Student-Teacher Ratio Discussion of Regression Results 291 Summary of Findings 295 Conclusion 296 APPENDI X 8.1 Regression Functions That Are Nonlinear in the Parameters 307 Assessing Studies Based on Multiple Regression Internal and External Validity Threats to Internal Validity 313 Threats to External Validity Threats to Internal Validity ofmultiple Regression Analysis 316 Omitted Variable Bias 316 Misspecification of the Functional Form ofthe Regression Function 319 Errors-in-Variables 319 Sample Selection

9 CONTE NTS xiii Simultaneous Causa li ty 324 Sources of Inconsistency of OLS Standard Errors Internal and External Validity When the Regression Is Used for Forecasting 327 Using Regression Models for Forecasting 327 Assessing the Va lidity of Regress ion Models for Forecasting Example: Test Scores and Class Size 329 External Validity 329 Internal Validity 336 Discuss ion and Implications S Conclusion 338 APPEND IX 9.1 The Massachusetts Elementary School Testing Data 344 PART THREE Further Topics in Regression Analysis 347 CHAPTER 10 Regression with Panel Data Panel Data 350 Example: Traffic Deaths and Alcohol Taxes Panel Data with Two Time Periods: "Before and After" Comparisons Fixed Effects Regression 356 The Fixed Effects Regression Model 356 Estimation and Inference 359 Application to Traffic Deaths Regression with Time Fixed Effects 361 Time Effects Only 36 1 Both En tity and Time Fixed Effects S The Fixed Effects Regression Assumptions and Standard Errors for Fixed Effects Regression 364 The Fixed Effects Regression Assumptions 364 Standard Errors for Fi xed Effects Regression Drunk Driving Laws and Traffic Deaths Conclusion 371 APPENDIX The State Traffic Fatality Data Set 378 APPENDIX 10.2 Standard Errors for Fixed Effects Regression with Serially Correlated Errors 379

10 xiv CONTENTS Regression with a Binary DependentVariable CHAPTER II 11.1 Binary Dependent Variables 385 The Linear Probability Model Probit and Logit Regression 389 Probit Regression 389 Logit Regression 394 Comparing the Linear Probability, Probit, and Logit Models 11.3 Nonlinear Least Squares Estimation 397 Maximum Likelihood Estimation 398 Measures of Fit Application to the Boston HMDA Data Binary Dependent Variables and the Linear Probability Model 396 Estimation and Inference in the Logit and Probit Models Summary 407 APPENDIX The Boston HMDA Data Set 415 APPENDIX Maximum Likelihood Estimati on 415 APPENDIX Other Limited Dependent Variable Models CHAPTER Instrumental Variables Regression 421 The IV Estimator w ith a Single Regressor and a Single Instrument 422 The IV Model and Assumptions 422 The Two Stage Least Squares Estimator 423 Why Does IV Regression Work? 424 The Sampling Distribution of the TSLS Estimator Application to the Demand for Cigarettes 430 The General IV Regression Model TSLS in the General IV Model Instrument Relevance and Exogeneity in the General IV Model The IV Regression Assumptions and Sampling Distribution ofthe TSLS Estimator 434 Inference Using the TSLS Estimator 437 Application to the Demand for Cigarettes 437 Checking Instrument Validity 439 Assumption # 1: Instrument Relevance 439 Assumption # 2: Ins trument Exogeneity 443 Application to the Demand for Cigarettes