Augmented Synthetic Control Method
The augmented synthetic control method adds an outcome-model bias correction, such as ridge regression, to synthetic control so treatment effects can be estimated when donor units poorly match the treated unit before treatment.
증강 합성통제법은 합성통제법에 릿지회귀 같은 결과모형 기반 편향보정을 더하여, 가중된 공여단위가 처치 이전 처치단위와 잘 맞지 않을 때도 처치효과를 추정하는 방법이다.
How do I estimate the effect of a policy on one region when no single comparison region is similar?
비슷한 비교 지역이 하나도 없을 때 한 지역에 대한 정책 효과를 어떻게 추정하나요?
The augmented synthetic control method builds a weighted combination of untreated units to approximate the treated unit's pre-treatment path and corrects remaining imbalance with an outcome model. It fits case studies with one or a few treated units and many pre-treatment periods where standard synthetic control fits poorly.
When to use Augmented Synthetic Control Method
- One or a few treated units, many untreated donor units and repeated outcome measures over time
- A policy or event affecting a single region, firm or country
- Standard synthetic control gives a poor pre-treatment fit
When to use something else
- Many treated units adopting at the same time with parallel trends plausible; use difference-in-differences
- Very few pre-treatment periods; use difference-in-differences or an event study
Assumptions and requirements
- No spillovers from treatment to donor units
- No anticipation of treatment
- Enough pre-treatment periods
- Donor pool unaffected by similar shocks
What the analysis reports
- Synthetic control weights
- Pre-treatment fit (RMSPE)
- Treated versus synthetic outcome plot
- Estimated average post-treatment effect with conformal or jackknife inference
- Placebo tests
How to report it
The augmented synthetic control estimate indicated an average post-treatment effect of [value] on [outcome] (p = [value]), with a pre-treatment RMSPE of [value].
Related methods
- Synthetic Control Method
- Difference-in-Differences
- Synthetic Difference-in-Differences
Run Augmented Synthetic Control Method on Quorant
- Data you need
- a table (CSV or Excel) with an outcome variable, a time variable, a unit (entity) ID and a treatment indicator
- Price
- 400 credits for one analysis up to 10,000 rows (complexity level 3); larger or wider data costs more. Optional methodologist review: 1800 credits.
- Checking
- Python results are cross-checked against R reference outputs in every deployment
- Before you pay
- A free pre-analysis diagnosis checks your data and gives an exact quote; nothing is deducted until you approve it.