mundlak

library(lme4)
library(gt)
library(modelsummary)
library(tidyverse)

# y dependent variable
# x regressor varies at the county-year level
# z regressor varies at the county level

url <- 'https://vincentarelbundock.github.io/Rdatasets/csv/plm/Crime.csv'
dat <- read_csv(url) %>%
       select(county, year,
              y = crmrte,
              x = density,
              z = pctmin) %>%
       group_by(county) %>%
       mutate(xbar = mean(x)) %>%
       ungroup
 
mod <- list()
mod[['Pooling']] <- lm(y ~ x + z, data = dat)
mod[['FE 1']] <- lm(y ~ x + factor(county), data = dat)
mod[['FE 2']] <- lm(y ~ I(x - xbar) - 1, data = dat)
mod[['MLM']] <- lmer(y ~ x + z + (1 | county), data = dat)
mod[['Mundlak']] <- lmer(y ~ I(x - xbar) + xbar + z + (1 | county), data = dat)

tab <- msummary(mod, fmt = '%.10f', coef_omit = 'county') %>%
       tab_style(cell_fill('pink'), cells_body('Mundlak', 7)) %>%
       tab_style(cell_fill('lightcyan'), cells_body('Mundlak', c(5, 11))) %>%
       tab_style(cell_fill('pink'), cells_body('FE 1', 3)) %>%
       tab_style(cell_fill('pink'), cells_body('FE 2', 7))

tab

tab %>% gtsave('mundlak.png')