SCSPILL

estimators.scspill.SCSPILL(config)

Spillover-aware synthetic control estimator (models sar and cd).

The default method="sar" fits the Bayesian spatial-autoregressive model of Sakaguchi and Tagawa (2026); method="cd" fits Cao and Dowd’s all-unit SCM with a known exposure structure and end-of-sample P-test inference (arXiv:1902.07343v3). Both return a :class:SCSPILLResults.

Parameters

Name	Type	Description	Default
config	SCSPILLConfig or dict	The estimation configuration; a dict is coerced into :class:scspill.config_models.SCSPILLConfig.	required

Examples

from scspill import SCSPILL
from scspill.data import load_california, load_california_cd

panel = load_california()
result = SCSPILL(
    {**panel.config_kwargs(), "m_iter": 2000, "burn": 1000, "seed": 42}
).fit()
result.att, result.att_ci
result.rho_hat, result.rho_ci
result.spillover_panel["Nevada"]

cd = SCSPILL(load_california_cd(51).config_kwargs()).fit()
cd.effects_panel.loc[1990, ["CA", "NV"]], cd.effect_ci["CA"]

Methods

Name	Description
fit	Estimate the selected model and return standardized results.

fit

estimators.scspill.SCSPILL.fit()

Estimate the selected model and return standardized results.

Runs the balance check and panel preparation shared by every model, the model’s own estimation (sar: the two-step sampler and the posterior effect sweep; cd: the all-unit SCM equations, the effects, and the P-test inference), and (optionally) the default plot panel.

Returns

Name	Type	Description
	SCSPILLResults	With the model’s fit in the slot named by method.

Raises

Name	Type	Description
	ScspillDataError	On invalid panel data or misaligned spatial weights.
	ScspillEstimationError	On an estimation failure (including an unidentified cd exposure structure).
	ScspillPlottingError	On plot generation failure (only when display_graphs).