Lecture notes for "Empirical Methods in Corporate Finance"

Below are the lecture notes for my PhD course on "Empirical Methods in Corporate Finance".

The first half of each 3 hour lecture covers an empirical methodology used in finance, while the second half of each lecture is comprised of students' discussions of three papers that apply the methodology discussed in the previous lecture.  The three assigned papers for each topic change from year to year.  Feel free to use these notes as a reference, but please provide appropriate acknowledgements.

In that regard, I would like to acknowledge the use of the following sources: (1) Wooldridge's 2010 textbook Econometric Analysis of Cross-Section and Panel Data, (2) Greene's 2011 textbook, Econometric Analysis, (3) Angrist and Pishke's 2009 book, Mostly Harmless Econometrics, (4) Roberts and Whited's 2011 working paper, "Endogeneity in Empirical Corporate Finance," and finally, (5) Michael Robert's PhD lecture notes for the same course in 2010.  Complete references can be found in the syllabus and at the start of each lecture.   

Please e-mail me if you have any questions or find any errors [which I'm sure there are many].  I also have PowerPoint versions of each lecture; if you would like a PowerPoint copy, please e-mail me.​

# 0 - Introduction

  • About me
  • Course objectives
  • Reading materials
  • Expected deliverables
  • Grading

#1 - Linear Regression, Part 1

  • The CEF and causality (very brief)
  • Linear OLS model
  • Multivariate estimation

#2 - Linear Regression, Part 2

  • Hypothesis testing
  • Miscellaneous issues (e.g., multicollinearity, interpreting interactions)

#3 - Causality

  • Why we care
  • Possible biases (omitted variables, measurement error, simultaneity)
  • Potential solutions

#4 - Panel Data

  • Why panel data is useful
  • Fixed effects
  • Random effects
  • First differences
  • Lagged y models

#5 - Instrumental Variables

  • ​Implementation and 2SLS
  • Weak instruments
  • Multiple IVs and overidentification 
  • Limitations of IV

#6 - Natural Experiment [Part 1]

  • Motivation and definitions
  • Understanding treatment effects
  • Two types of simple differences
  • Difference-in-differences

07 -- Natural Experiment [Part 2]*

  • Additional issues with diff-in-diffs
  • How to handle multiple events
  • Stacked diff-in-diffs
  • Falsification tests
  • Triple differences

#8 - Regression Discontinuity

  • Basic idea of RD
  • Sharp versus fuzzy discontinuities
  • Checks on internal validity
  • Heterogeneous effects
  • External validity

#9 - Common Limitations & Errors

  • The data we use is almost never perfect (e.g., survivorship issues)
  • Hypothesis testing mistakes
  • How to (and not to) control for unobserved heterogeneity

#10 - Matching & Selection Models

  • Introduction to matching
  • How to do matching
  • Practical considerations
  • Testing the assumptions
  • Key weaknesses 
  • Heckman selection model

#11 - Standard Errors & Misc.

  • Classical vs. robust standard errors
  • Clustered standard errors
  • Limited dependent variables

​* I have now updated the "Natural Experiment" slides to reflect recent work that discusses how one should best handle multiple events in a diff-in-diffs setting.  I personally think the stacked diff-in-diffs approach developed in Gormley and Matsa (2011, 2016) is best, which can now be easily implemented in Stata via the STACKDID command (Thanks Jacob Triplett!). However, Michael Wittry has also updated the above slides for his own PhD course at Ohio State and was kind enough to share them.  They can be found by clicking here.  As you'll see, his slides combine lectures #6 and #7 above and add new discussion of how to handle multiple events (e.g., see slides 85-115 of his slide deck).  His slides follow the excellent guidance provided in Baker, Larcker, and Wang (JFE 2022).  If you would like the PPT version of Michael's slides, please reach out directly to Michael at [email protected].  Thanks for sharing Michael!