Perform a Donut FE analysis — donut

Performs a single FE-models with a given data set. The independent variable and the population can be adjusted.

Usage

donut_analysis(
  dist,
  ds,
  dep_var,
  indep_vars,
  weights = NULL,
  fe = "id",
  dist_var = "dist_km",
  se = "cluster",
  bootstrap = FALSE,
  B = 9999,
  lat = "lat",
  lon = "lon",
  excl_largest_fe = 0,
  excl_inner_r = 0,
  treatment_label = "Lives close to landfill",
  ...
)

Arguments

dist: A vector with two distances: the inner and the outer. The inner distance refers to the maximum distance for treatment. The outer distance refers to the overall population: it is the cut-off value.
ds: The dataset.
dep_var: The dependent variable in the regression
indep_vars: a character vector of independent variables. Must be part of ds.
weights: a numeric vector for weights. Must be part of ds.
fe: a character vector of Fixed Effects variables. Must be part of ds.
dist_var: A character referring to the distance parameter. Defaults to dist_km.
se: is a character, defaulting to cluster. It shows how to adjust standard errors: By clustering them by the fe param (cluster), no adjustment, (basic), or with Conley (conley). All regressions rely on fixest.
bootstrap: a boolean: should the standard errors for dist be bootstrapped?
B: the number of bootstraps
lat: Latitude; needed for conley-standard errors
lon: Longitude; needed for conley-standard errors
excl_largest_fe: Numeric: how many of the FE with the largest number of observations should be excluded? Defaults to 0.
excl_inner_r: Numeric. If the observations that are closest to the fe should be omitted: in which radius should they be omitted? Defaults to 0.
treatment_label: Character. The label for the treatment variable created for the combination of inner and outer radius. Defaults to "Lives close to landfill"
...: Additional arguments

Value

A single donut_model object with class fixest. The donut_model has additional list element which contain the inner and outer radii and additional info. It can be plotted with the plot_significance function, and showed as table with donut_summary.

Examples

data(donut_data)
model1 <-
donut_analysis(dist = c(5, 20),
               ds = donut_data,
               dep_var = "wealth_index",
               indep_vars = "age",
               fe = "id")
model2 <-
  donut_analysis(dist = c(8, 50),
                 ds = donut_data,
                 dep_var = "wealth_index",
                 indep_vars = c("age", "male"),
                 fe = "id",
                 dist_var = "dist_km",
                 excl_largest_fe = 2,
                 excl_inner_r = 1)
model_basic <-
  donut_analysis(dist = c(8, 50),
                 ds = donut_data,
                 dep_var = "wealth_index",
                 indep_vars = c("age", "male"),
                 fe = "id",
                 dist_var = "dist_km",
                 se = "basic")