run_synth_bsts: Synthetic Control via BSTS (CausalImpact)

In bivarhr: Bivariate Hurdle Regression with Bayesian Model Averaging

Synthetic Control via BSTS (CausalImpact)

Description

Builds a simple synthetic-control-style analysis using CausalImpact/BSTS for either I or C as the outcome, with treatment defined endogenously by a high level of a chosen control variable.

Usage

run_synth_bsts(
  DT,
  outcome = c("I", "C"),
  control_var,
  seed = 123,
  dir_csv = NULL
)

Arguments

DT	A data.frame or data.table containing at least: I, C: outcome candidates (counts or rates). EconCycle, PopDensity, Epidemics, Climate, War, t_norm: predictors used to build the synthetic control. The column named in control_var, used to define the treated period.
outcome	Character; which outcome series to use as the response, one of "I" or "C".
control_var	Character scalar; name of a column in DT whose high values define the treated period (e.g., intensity of some intervention or shock proxy).
seed	Integer; random seed for reproducibility of the BSTS fit.
dir_csv	Character scalar or NULL; directory where the effect summary CSV is written. If NULL (default), nothing is written to disk.

Details

This function requires the CausalImpact package (listed under Suggests); an informative error is raised at call time if it is not installed.

The function:

1. Selects the outcome series y <- DT[[outcome]].

2. Builds the predictor matrix from EconCycle, PopDensity, Epidemics, Climate, War, and t_norm.

3. Uses control_var to define a treated period as observations where control_var is in the top third (>= 2/3 quantile). If fewer than 5 treated observations are found, the function returns NULL.

4. Sets the intervention start time t0 as one period before the first treated index (with a minimum of 10 observations in the pre-period). The pre- and post-intervention windows are: pre.period = c(1, t0) and post.period = c(t0 + 1, length(y)).

5. Calls CausalImpact::CausalImpact() on the combined cbind(y, preds) matrix, with model.args = list(nseasons = 1).

From the resulting impact object, the function extracts the average absolute and relative effects from impact$summary and stores them in a small summary table with two rows: "abs_effect_mean" and "rel_effect_mean".

When dir_csv is supplied, a CSV file named "causalimpact_<control_var>_on_<outcome>.csv" is written to that directory. If CausalImpact() fails, the function returns NULL.

Value

On success, a list with components:

• impact: the full CausalImpact object.

• summary: a data.frame with the mean absolute and relative effects.

If the treated period is too short or the model fit fails, the function returns NULL.

Examples

# This example runs only when 'CausalImpact' is installed.
if (requireNamespace("CausalImpact", quietly = TRUE)) {
  DT <- data.frame(
    I = rpois(30, lambda = 10),
    C = rpois(30, lambda = 8),
    EconCycle = rnorm(30),
    PopDensity = rnorm(30),
    Epidemics = rnorm(30),
    Climate = rnorm(30),
    War = rnorm(30),
    t_norm = seq(-1, 1, length.out = 30)
  )

  res_I <- run_synth_bsts(DT, outcome = "I", control_var = "War", seed = 123)
  if (!is.null(res_I)) {
    print(res_I$summary)
  }
}

bivarhr documentation built on July 7, 2026, 1:06 a.m.