In evanjflack/bacondecomp: Goodman-Bacon Decomposition

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

bacondecomp

bacondecomp is a package with tools to perform the Goodman-Bacon decomposition for differences-in-differences with variation in treatment timing. The decomposition can be done with and without time-varying covariates.

Installation

You can install bacondecomp 0.1.1 from CRAN:

install.packages("bacondecomp")

You can install the development version of bacondecomp from GitHub:

library(devtools)
install_github("evanjflack/bacondecomp")

Functions

• bacon(): calculates all 2x2 differences-in-differences estimates and weights for the Bacon-Goodman decomposition.

Data

• math_refom: Aggregated data from Goodman (2019, JOLE)
• castle: Data from Cheng and Hoekstra (2013, JHR)
• divorce: Data from Stevenson and Wolfers (2006, QJE)

Example

This is a basic example which shows you how to use the bacon() function to decompose the two-way fixed effects estimate of the effect of an education reform on future earnings following Goodman (2019, JOLE).

library(bacondecomp)
df_bacon <- bacon(incearn_ln ~ reform_math,
                  data = bacondecomp::math_reform,
                  id_var = "state",
                  time_var = "class")

# All 2x2 Comparisons
head(df_bacon)

# Summary of Early vs. Later, Later vs. Earlier, and Treated vs. Untreated
bacon_summary(df_bacon)

library(ggplot2)

ggplot(df_bacon) +
  aes(x = weight, y = estimate, shape = factor(type)) +
  geom_point() +
  geom_hline(yintercept = 0) + 
  theme_minimal() +
  labs(x = "Weight", y = "Estimate", shape = "Type")

References

Goodman-Bacon, Andrew. 2018. "Difference-in-Differences with Variation in Treatment Timing." National Bureau of Economic Research Working Paper Series No. 25018. doi: 10.3386/w25018.

Paper Link

evanjflack/bacondecomp documentation built on Sept. 19, 2021, 8:53 a.m.