Recomputes the overall estimate dropping each of the top sectors (by |Rotemberg weight|) one at a time, to see whether identification hinges on a single shock.
Usage
ssb_loo(
design,
top = 5,
se = c("none", "iid", "ehw", "cluster", "akm", "akm0"),
level = 0.95
)Arguments
- design
An [ssb_design()] object.
- top
Number of top-weight sectors to leave out in turn.
- se
Standard-error method for a confidence interval on each leave-one-out estimate: `"none"` (default; point estimates only, the original behaviour) or one of `"iid"`, `"ehw"`, `"cluster"`, `"akm"`, `"akm0"` (each re-estimated on the reduced design via [ssb_estimate()]). With a CI you can read whether the estimate still excludes 0 after dropping the most influential shock; [ssb_plot_loo()] then draws the intervals.
- level
Confidence level for the interval when `se` is not `"none"`.
Value
A `data.frame` with the dropped `sector`, its `alpha`, and the `beta_drop` obtained without it (plus the full-sample `beta_hat` attribute). When `se` is not `"none"` it also has `conf.low`/`conf.high` columns and `se_method`/`level` attributes.
Examples
sim <- ssb_simulate(n_loc = 80, n_sec = 10, seed = 1)
d <- ssb_design(sim$data, sim$shares, sim$shocks, exogenous = "share")
ssb_loo(d, top = 5)
#> <ssBartik leave-one-out> overall beta = 1.4468
#> sector alpha beta_drop
#> 6 0.5850 1.40
#> 3 0.1736 1.44
#> 8 0.1109 1.57
#> 4 0.0875 1.42
#> 9 0.0165 1.42