In s3alfisc/fwildclusterboot: Fast Wild Cluster Bootstrap Inference for Linear Models

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

fwildclusterboot

r badger::badge_cran_download("fwildclusterboot", "grand-total", "blue") r badger::badge_cran_download("fwildclusterboot", "last-month", "green")

The {fwildclusterboot} package implements multiple fast wild cluster bootstrap algorithms as developed in Roodman et al (2019) and MacKinnon, Nielsen & Webb (2022).

Via the JuliaConnectoR, {fwildclusterboot} further ports functionality of WildBootTests.jl - which provides an even faster implementation of the wild cluster bootstrap for OLS and supports the WRE bootstrap for IV and tests of multiple joint hypotheses.

The package's central function is boottest(). It allows to test univariate hypotheses using a wild cluster bootstrap at extreme speed: via the 'fast' algorithm, it is possible to run a wild cluster bootstrap with $B = 100.000$ iterations in less than a second!

library(ggplot2)
df <- readRDS("C:/Users/alexa/Dropbox/R package development/fwildclusterboot develop/benchmarks/r_comparions.rds")

df$B <- factor(df$B, levels = c("10K", "100K"))
df$N_G <- factor(df$N_G, levels = c("N_G = 20", "N_G = 50", "N_G = 100", "N_G = 500", "N_G = 1000"))
df$`boot algo` <- df$type

ggplot(data = df, aes(x = B, y = time, color = `boot algo`)) + 
  facet_wrap(~N_G, nrow = 1) + 
  geom_point() + 
scale_y_continuous(trans='log10') + 
  labs(title = "Benchmarks", caption = "N = 10_000, k = 20 covariates and one cluster of dimension N_G (3 iterations each, median runtime is plotted).")+
  #theme_bw() + 
  xlab("Bootstrap iterations") + 
  ylab("time in seconds, log scale") + 
  theme_bw()

{fwildclusterboot} supports the following features:

• The wild bootstrap for OLS (Wu 1986).
• The wild cluster bootstrap for OLS (Cameron, Gelbach & Miller 2008, Roodman et al, 2019).
• Multiple new versions of the wild cluster bootstrap as described in MacKinnon, Nielsen & Webb (2022), including the WCR13 (WCR-V), WCR31 (WCR-S), WCR33 (WCR-B), WCU13 (WCU-V), WCU31 (WCU-S) and WCU33 (WCU-B).
• The subcluster bootstrap (MacKinnon and Webb 2018).
• Confidence intervals formed by inverting the test and iteratively searching for bounds.
• Multiway clustering.
• One-way fixed effects.

Additional features are provided through WildBootTests.jl:

• Highly optimized versions of the '11' and '31' wild cluster bootstrap variants
• A highly optimized version of the Wild Restricted Efficient bootstrap (WRE) for IV/2SLS/LIML (Davidson & MacKinnon, 2010).
• Arbitrary and multiple linear hypotheses in the parameters.

{fwildclusterboot} supports the following models:

• OLS: lm (from stats), fixest (from fixest), felm from (lfe)
• IV: ivreg (from ivreg).

Installation

You can install compiled versions of{fwildclusterboot} from CRAN (compiled), R-universe (compiled) or github by following one of the steps below:

# from CRAN 
install.packages("fwildclusterboot")

# from r-universe (windows & mac, compiled R > 4.0 required)
install.packages('fwildclusterboot', repos ='https://s3alfisc.r-universe.dev')
# dev version from github
# note: installation requires Rtools
library(devtools)
install_github("s3alfisc/fwildclusterboot")

The boottest() function

For a longer introduction to {fwildclusterboot}, take a look at the vignette.

library(fwildclusterboot)

# set seed via dqset.seed for engine = "R" & Rademacher, Webb & Normal weights
dqrng::dqset.seed(2352342)
# set 'familiar' seed for all other algorithms and weight types 
set.seed(23325)

data(voters)

# fit the model via fixest::feols(), lfe::felm() or stats::lm()
lm_fit <- lm(proposition_vote ~ treatment  + log_income + as.factor(Q1_immigration) + as.factor(Q2_defense), data = voters)
# bootstrap inference via boottest()
lm_boot <- boottest(lm_fit, clustid = c("group_id1"), B = 9999, param = "treatment")
summary(lm_boot)

Citation

If you are in R, you can simply run the following command to get the BibTeX citation for {fwildclusterboot}:

citation("fwildclusterboot")

Alternatively, if you prefer to cite the "Fast & Wild" paper by Roodman et al, it would be great if you mentioned {fwildclusterboot} in a footnote =) !

s3alfisc/fwildclusterboot documentation built on Sept. 17, 2023, 5:55 a.m.