mdd_event_study: Event study

In MatthieuStigler/multiDiff: Multi-year diff-diff and FE decompositions

Event study

Description

Event study

Usage

mdd_event_study(
  mdd_dat,
  trim_low = NULL,
  trim_high = NULL,
  time.omit = -1,
  weights = NULL,
  cluster = NULL,
  ...
)

## S3 method for class 'mdd_event_study'
plot(x, ...)

Arguments

mdd_dat	An object of class mdd_dat created by mdd_data_format
trim_low, trim_high	Upper/lower bound on parameters to include
time.omit	Which is the base year omitted in the analysis?
weights	Weight argument passed to the underlying feols function
cluster	Cluster argument passed to the underlying feols function
...	passed to feols for mdd_event_study
x	the ES object

Details

This function uses lags of pre/post treatment, omitting the lag -1 (can be changed with time.omit). The interpretation of the event-study coefficient is then the following: each coefficient represents the value of a DiD relating the treated/control difference at relative year -3, -2, 1, 2 to the treated/control difference at relative year -1.

See Also

mdd_test_pre_trend_event to run a parallel trend assumption

Examples

## simulate and format data
DID_dat <- sim_dat_common(timing_treatment = 5:10, as_mdd = TRUE)

## Estimate DiD
mdd_DD_simple(DID_dat)

## estimate ES
ES_out <- mdd_event_study(DID_dat)
ES_out
summary(ES_out)
plot(ES_out)

## compare to year-t- base year diff-diffs:
means_manu <- DID_dat |>
  mdd_group_means() |>
  as.data.frame() |>
  reshape(idvar = "Time",timevar = ".group",
          direction = "wide")
diff_by_period <- means_manu$y.treated - means_manu$y.control
  all.equal((diff_by_period - diff_by_period[5])[-5],
            coef(ES_out), check.attributes= FALSE)

MatthieuStigler/multiDiff documentation built on Oct. 21, 2023, 8:13 a.m.