Skip to content

R API Reference

distributed_lags_model()

Estimate a single-outcome distributed lag model.

Usage

distributed_lags_model(
  data,
  exposure_data,
  from_rt,
  to_rt,
  outcome,
  exposure,
  unit,
  time,
  covariates = NULL,
  addl_fes = NULL,
  ref_period = -1,
  weights = NULL,
  dd = FALSE,
  n = 2,
  dict = NULL
)

Arguments

Argument Description
data Data frame with unit, time, outcome, and covariates
exposure_data Data frame with unit, time, and exposure variable
from_rt Starting relative period (negative integer)
to_rt Ending relative period (positive integer)
outcome String: name of the outcome variable
exposure String: name of the exposure variable
unit String: name of the unit identifier
time String: name of the time variable
covariates Character vector of covariate names (default: NULL)
addl_fes Character vector of additional FE variables (default: NULL)
ref_period Reference period (default: −1)
weights String: name of weight variable (default: NULL)
dd Logical: include DD companion estimate in plot (default: FALSE)
n Integer: number of digits for DD companion (default: 2)
dict Named vector for renaming axes in the plot (default: NULL)

Return Value

A list with:

Element Type Description
betas data.frame Columns: time_to_event, coef, se
plot ggplot Event-study plot with confidence intervals
model fixest The estimated feols model object
vcov matrix Variance-covariance matrix of gamma coefficients
data_periods_included vector Calendar periods included in estimation
fmla_str string Formula string used in feols
from_rt numeric From period (echo)
to_rt numeric To period (echo)
exposure string Exposure variable name (echo)
outcome string Outcome variable name (echo)

distributed_lags_models()

Estimate DLMs for multiple outcomes in a single call.

Usage

distributed_lags_models(
  data, exposure_data, from_rt, to_rt,
  outcomes, exposure, unit, time, ...
)

Takes the same arguments as distributed_lags_model(), except outcomes (plural) is a character vector. Returns a list of model results, one per outcome.

generate_data()

Generate a balanced panel with staggered treatment for testing.

Usage

generate_data(seed = 1234, n_groups = 676, n_times = 20, treat_prob = 0.4)

Return Value

A data frame with columns: group, time, treat, treatment_time, years_to_treatment, post, outcome.

standard_twfe_for_comparison()

Estimate a standard binned-endpoint TWFE event study for comparison with DLM results.

Usage

standard_twfe_for_comparison(
  data, from_rt, to_rt, outcome, time, unit,
  time_to_treatment, treat, covariates = NULL,
  ref_period = -1, weights = NULL
)

Helper Functions

Function Description
revcumsum(x) Reverse cumulative sum of a vector
secumsum(cov) SE of forward cumulative sum from a VCV matrix
serevcumsum(cov) SE of reverse cumulative sum from a VCV matrix