Doubly robust DiD over a staggered adoption design: the ATT(g,t) table

One doubly robust estimate per (cohort, event-time) cell the panel can support, each between the base period g - 1 and period g + e, against the never-treated units. Pre-treatment cells are computed on identical footing and returned rather than suppressed.

Usage

Drsta(y, D, unit, time, cohort, X = NULL)

Arguments

y

Outcome, long format, one entry per unit-period.

D

Treatment indicator, carried for the interface.

unit, time

Unit and period identifiers.

cohort

First treated period of the row's unit; 0 or Inf marks a never-treated unit.

X

Optional baseline covariates, one row per unit-period.

Value

List with estimate, cohorts, event_time, att (row-major cohort x event-time), n_cells, n_post, n.

References

Callaway, B. and Sant'Anna, P. H. C. (2021). Journal of Econometrics 225(2), 200-230. Sant'Anna, P. H. C. and Zhao, J. (2020). Journal of Econometrics 219(1), 101-122, equation (2.6). Roth, J. and Sant'Anna, P. H. C. (2023). Journal of Political Economy Microeconomics 1(4), 669-709.

Examples

set.seed(1)
r <- Drsta(y = rnorm(10), D = rbinom(10, 1, 0.5), unit = rnorm(10), time = sort(runif(10)),
  cohort = rnorm(10))
TRUE
#> [1] TRUE