Skip to content

Latest commit

 

History

History
54 lines (42 loc) · 2.15 KB

README.md

File metadata and controls

54 lines (42 loc) · 2.15 KB
Raw

covid-impact

Repository for Chernozhukov, Kasahara, and Schrimpf "Causal Impact of Masks, Policies, Behavior on Early Covid-19 Pandemic in the U.S."

Instructions

Install Packages

We use many R packages. An easy way to install all the dependencies is with

First install renv.

if (!requireNamespace("remotes"))
  install.packages("remotes")

remotes::install_github("rstudio/renv")

You may also need to install pandoc. One way to do so is with the installr package.

Clone this git repository. Make sure R's working directory is inside the cloned git repository and run the following:

loadNamespace("renv") # library(renv) will overwrite some base functions and break things
rootdir <- system("git rev-parse --show-toplevel", intern=TRUE)[1]
setwd(rootdir) # this may not be needed, I don't completely know how renv works
renv::restore()

Run the code

Code to generate the main regression results is in rmd/regressionsWithPlots.Rmd.

Code for the counterfactuals is in rmd/counterfactuals.Rmd.

In either case, you can generate the latex tables, pdf figures, and html output by running

library(rmarkdown)
rootdir <- system("git rev-parse --show-toplevel", intern=TRUE)[1]
rmarkdown::render(paste(rootdir,"cases_and_policies/rmd/regressionsWithPlots.Rmd", sep="/"))
rmarkdown::render(paste(rootdir,"cases_and_policies/rmd/counterfactuals.Rmd", sep="/"))

Caveats

The usage of renv was added after the paper was published, so versions of packages in the renv environment may not match those used in the publication. In particular, it appears that plm has had some changes in behavior (or another package has changes and interferes with plm). I think I have updated the code to work, but there might still be problems.

While updating the code to work with updated dependencies, a bug in the code for the PI->B table was unknowingly corrected. The version of the table in the published paper incorrectly has month included as an integer. The corrected code includes month dummies. Other regressions include month dummies in both the published version and the updated code. Thanks to Philippe Lemoine for pointing out this discrepency.