Knowingo
Try 7 days free
The cover of Mostly Harmless Econometrics

Mostly Harmless Econometrics

Joshua D. Angrist, Jörn Steffen Pischke

Which econometric methods are essential for drawing causal conclusions from data, and which exotic ones can you safely skip?

Get the app Read on the web

At a glance

Author
Joshua D. Angrist, Jörn Steffen Pischke

Why read it

  • You will learn the core tools: regression to control for confounding variables, instrumental variables, and differences in differences
  • You will see why a randomized experiment is the ideal benchmark, and how to judge any study by how close it comes
  • You will be able to judge whether a set of regression estimates has a causal interpretation

Who it is for

  • A doctoral student in economics building a practical toolkit
  • A public health researcher estimating the effect of a program
  • A policy analyst who must read empirical papers critically

Contents

  1. PREFACE
  2. ORGANIZATION OF THIS BOOK
  3. Part I - Preliminaries
  4. Chapter 1 - Questions about Questions
  5. Chapter 2 - The Experimental Ideal
  6. 2.1 The Selection Problem
  7. 2.2 Random Assignment Solves the Selection Problem
  8. 2.3 Regression Analysis of Experiments
  9. Part II - The Core
  10. Chapter 3 - Making Regression Make Sense
  11. 3.1 Regression Fundamentals
  12. 3.1.1 Economic Relationships and the Conditional Expectation Function
  13. 3.1.2 Linear Regression and the CEF
  14. 3.1.3 Asymptotic OLS Inference
  15. 3.1.4 Saturated Models, Main Effects, and Other Regression Talk
  16. 3.2 Regression and Causality
  17. 3.2.1 The Conditional Independence Assumption
  18. 3.2.2 The Omitted Variables Bias Formula
  19. 3.2.3 Bad Control
  20. 3.3 Heterogeneity and Nonlinearity
  21. 3.3.1 Regression Meets Matching
  22. 3.3.2 Control for Covariates Using the Propensity Score
  23. 3.3.3 Propensity Score Methods versus Regression
  24. 3.4 Regression Details
  25. 3.4.1 Weighting Regression
  26. 3.4.2 Limited Dependent Variables and Marginal Effects
  27. 3.4.3 Why Is Regression Called Regression, and What Does Regression to the Mean Mean?
  28. 3.5 Appendix: Derivation of the Average Derivative Weighting Function
  29. Chapter 4 - Instrumental Variables in Action: Sometimes You Get What You Need
  30. 4.1 IV and Causality
  31. 4.1.1 Two-Stage Least Squares
  32. 4.1.2 The Wald Estimator
  33. 4.1.3 Grouped Data and 2SLS
  34. 4.2 Asymptotic 2SLS Inference
  35. 4.2.1 The Asymptotic Distribution of the 2SLS Coefficient Vector
  36. 4.2.2 Overidentification and the 2SLS Minimand
  37. 4.3 Two-Sample IV and Split-Sample IV
  38. 4.4 IV with Heterogeneous Potential Outcomes
  39. 4.4.1 Local Average Treatment Effects
  40. 4.4.2 The Compliant Subpopulation
  41. 4.4.3 IV in Randomized Trials
  42. 4.4.4 Counting and Characterizing Compliers
  43. 4.5 Generalizing LATE
  44. 4.5.1 LATE with Multiple Instruments
  45. 4.5.2 Covariates in the Heterogeneous Effects Model
  46. 4.5.3 Average Causal Response with Variable Treatment Intensity
  47. 4.6 IV Details
  48. 4.6.1 2SLS Mistakes
  49. 4.6.2 Peer Effects
  50. 4.6.3 Limited Dependent Variables Reprise
  51. 4.6.4 The Bias of 2SLS
  52. 4.7 Appendix
  53. Chapter 5 - Parallel Worlds: Fixed Effects, Differences-in-Differences, and ...
  54. 5.1 Individual Fixed Effects
  55. 5.2 Differences-in-Differences: Pre and Post, Treatment and Control
  56. 5.2.1 Regression DD
  57. 5.3 Fixed Effects versus Lagged Dependent Variables
  58. 5.4 Appendix: More on Fixed Effects and Lagged Dependent Variables
  59. Part III - Extensions
  60. Chapter 6 - Getting a Little Jumpy: Regression Discontinuity Designs

More books like this