Synthetic Control Method (SCM)
Synthetic Control Method for Comparative Case Studies · Also known as: SCM, synthetic control, synth estimator, Abadie-Diamond-Hainmueller method
The Synthetic Control Method estimates the causal effect of a treatment or policy on a single treated unit by constructing a weighted combination of untreated units — the synthetic control — that closely resembles the treated unit before the intervention. The gap between the treated unit and its synthetic counterpart after the intervention is the estimated treatment effect.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
+35 more
When to use it
Use the synthetic control method when a single or very few aggregate units (countries, states, regions, firms) are treated by a well-defined intervention, a long pre-treatment panel of outcomes is available for a set of similar untreated units, and the intervention does not affect all units simultaneously. It is especially valuable when no natural comparison unit exists and when difference-in-differences assumptions (parallel trends) are too restrictive. Do not use it when the donor pool is very small (fewer than five donors), when pre-treatment fit cannot be achieved because no weighted combination of donors tracks the treated unit, when the treatment unit is an outlier with no comparable donors, or when the outcome series is too short to credibly establish pre-treatment balance.
Strengths & limitations
- Provides a transparent, data-driven counterfactual without requiring the parallel-trends assumption to hold for a single comparison group.
- Non-negative weights summing to one prevent extrapolation beyond the convex hull of the donor pool, making the synthetic control interpretable.
- Visual display of the treated and synthetic control paths makes the quality of fit and the post-intervention gap immediately apparent to readers.
- Placebo-based inference does not rely on large-sample asymptotics and is valid even when there is only one treated unit.
- Pre-treatment fit quality is directly observable and serves as a built-in diagnostic for the credibility of the counterfactual.
- The method is designed for aggregate units; it is not suited to individual-level microdata without substantial aggregation.
- Achieving a good pre-treatment fit requires many pre-treatment periods; a short baseline undermines the credibility of the synthetic control.
- With a small donor pool, the distribution of placebo gaps is coarse and inference power is limited.
- The method does not straightforwardly generalize to multiple simultaneously treated units without extensions such as the generalized synthetic control.
Frequently asked
How is the synthetic control different from difference-in-differences?
DiD compares a treated group to a single (or equally-weighted) control group under a parallel-trends assumption. The synthetic control constructs an optimally-weighted combination of multiple donors that best matches the treated unit before the intervention, making the counterfactual more flexible and transparent. It is especially suited to one treated unit; DiD is better when many treated units are available.
How do I assess whether the synthetic control is credible?
Inspect the pre-treatment fit: plot both series and compute the root mean squared prediction error (RMSPE) in the pre-period. A low RMSPE relative to the donor pool indicates a credible counterfactual. If the pre-treatment gap is as large as the post-treatment gap, the result is not informative.
How is statistical significance tested without a p-value from a standard regression?
Significance is assessed through placebo (permutation) tests: the synthetic control procedure is applied to each donor unit as if it were treated. If the treated unit's post-intervention RMSPE ratio (post/pre) is much larger than those of the placebos, this provides evidence that the observed gap is unlikely to arise by chance.
Can I use the synthetic control with many treated units?
The original method targets a single treated unit. For multiple treated units or staggered adoption, extensions such as the generalized synthetic control (Xu, 2017) or the augmented synthetic control (Ben-Michael et al., 2021) are recommended.
What is a good donor pool size?
A donor pool of at least ten to twenty untreated units is generally advisable to allow meaningful optimization and credible placebo inference. Fewer than five donors severely limits both pre-treatment fit options and the power of permutation tests.
Sources
- Abadie, A., Diamond, A., & Hainmueller, J. (2010). Synthetic Control Methods for Comparative Case Studies: Estimating the Effect of California's Tobacco Control Program. Journal of the American Statistical Association, 105(490), 493-505. DOI: 10.1198/jasa.2009.ap08746 ↗
- Abadie, A., & Gardeazabal, J. (2003). The Economic Costs of Conflict: A Case Study of the Basque Country. American Economic Review, 93(1), 113-132. DOI: 10.1257/000282803321455188 ↗
How to cite this page
ScholarGate. (2026, June 3). Synthetic Control Method for Comparative Case Studies. ScholarGate. https://scholargate.app/en/causal-inference/synthetic-control-method
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Causal Impact AnalysisCausal inference↔ compare
- Difference-in-DifferencesEconometrics↔ compare
- Instrumental Variables in Health ResearchHealth Economics↔ compare
- Propensity Score MatchingResearch Statistics↔ compare