The Augmented Synthetic Control Method: The Kansas Tax Cuts

Downloads

Each dataset is available as a labeled Stata .dta and its source file.

⇩ Download all data (ZIP)stata_codebook.do

Run stata_codebook.do in Stata once to attach long-form per-variable notes to the .dta files.

Load directly in code

Every file loads straight from GitHub (raw URLs). Swap the file name to load any dataset.

Stata

* Stata 14+ : `use` reads an https URL directly
global BASE "https://raw.githubusercontent.com/cmg777/starter-academic-v501/master/content/post/r_augsynth/data/"
use "${BASE}kansas.dta", clear
describe
notes

Python

!pip install -q pyreadstat
import pandas as pd
BASE = "https://raw.githubusercontent.com/cmg777/starter-academic-v501/master/content/post/r_augsynth/data/"
df = pd.read_stata(BASE + "kansas.dta")

# load every dataset at once
files = ["kansas"]
data = {f: pd.read_stata(BASE + f + ".dta") for f in files}

# pyreadstat (richest metadata) reads LOCAL files -> download first
import pyreadstat, urllib.request
urllib.request.urlretrieve(BASE + "kansas.dta", "kansas.dta")
df, meta = pyreadstat.read_dta("kansas.dta")

Copy and paste this snippet in Google Colab app. https://colab.research.google.com/notebooks/empty.ipynb

R

# R : haven::read_dta auto-downloads an https URL
library(haven)
BASE <- "https://raw.githubusercontent.com/cmg777/starter-academic-v501/master/content/post/r_augsynth/data/"
df <- read_dta(paste0(BASE, "kansas.dta"))

Overview & sources

Companion data for a beginner-friendly R tutorial on the Augmented Synthetic Control Method (ASCM) of Ben-Michael, Feller & Rothstein (2021), estimating the effect of the May 2012 Kansas personal-income-tax cut on log gross state product (GSP) per capita. The dataset is the kansas object shipped with the augsynth R package: a balanced panel of 50 U.S. states observed every quarter from 1990 Q1 to 2016 Q1 (105 quarters per state, 5,250 rows). Kansas (FIPS 20) is the single treated unit, switched on from 2012 Q2 (year_qtr = 2012.25) onward; the other 49 states form the donor pool. The outcome is lngdpcapita; per-capita revenue, wage, establishment and employment series serve as auxiliary covariates. The tutorial moves from classic SCM through ridge augmentation and covariate balancing, then compares four ways to do inference.

Data sources

Cite this data

Please cite this dataset as follows.

APA

Mendez, C. (2026). The Augmented Synthetic Control Method: A Beginner's Tutorial with the Kansas Tax Cuts [Data set]. https://carlos-mendez.org/post/r_augsynth/

Ben-Michael, E., Feller, A., & Rothstein, J. (2021). The Augmented Synthetic Control Method. Journal of the American Statistical Association, 116(536), 1789–1803. https://doi.org/10.1080/01621459.2021.1929245

BibTeX

@misc{mendez2026raugsynth,
  author       = {Mendez, Carlos},
  title        = {The Augmented Synthetic Control Method: A Beginner's Tutorial with the Kansas Tax Cuts},
  year         = {2026},
  howpublished = {\url{https://carlos-mendez.org/post/r_augsynth/}},
  note         = {Data set}
}

@article{benmichael2021augmented,
  author  = {Ben-Michael, Eli and Feller, Avi and Rothstein, Jesse},
  title   = {The Augmented Synthetic Control Method},
  journal = {Journal of the American Statistical Association},
  volume  = {116}, number = {536}, pages = {1789--1803}, year = {2021},
  doi     = {10.1080/01621459.2021.1929245}
}

Variable explorer search & filter all 32 variables

Type to filter by name or label, or use the chips to filter by type. Each row shows a mini distribution. Click a header to sort.

Cross-file variable index

Which file each variable appears in (● = present).

Construction & formulas

The estimand is the ATT — actual minus synthetic, for Kansas, after 2012 Q2 — on the outcome lngdpcapita. Write X1 for Kansas's pre-treatment outcome vector and X0 for the donors' matching matrix.

• Classic SCM weights (progfunc = "None"): γ̂_scm = argmin_γ ‖X1 − X0'γ‖₂² subject to Σ γ_i = 1, γ_i ≥ 0 — a non-negative, sum-to-one (convex) blend of donors.
• Per-quarter effect: τ̂_t = Y_1t − Σ_i γ̂_i Y_it for t > T₀ (actual minus synthetic).
• Ridge-augmented counterfactual (progfunc = "Ridge"): Ŷ_aug(0) = Σ_i γ̂_i Y_iT + (m̂_1T − Σ_i γ̂_i m̂_iT) — SCM plus a bias correction from a ridge outcome model m̂ of post-period outcomes on lagged outcomes, penalty λ‖η‖₂² (λ chosen by leave-one-pre-period-out CV, one-SE rule).
• Covariate ASCM: list covariates after a | in the formula (lngdpcapita ~ treated | log(revstatecapita) + … + emplvlcapita); both lagged outcomes and covariates enter the SCM balancing problem and the ridge model.

Per-capita series are constructed by dividing a state-quarter total by population (popestimate); lngdp/lngdpcapita are natural logs of gdp/gdpcapita. year_qtr encodes the quarter as year + (qtr−1)/4 (e.g. 2012 Q2 = 2012.25).

The datasets

Switch datasets with the tabs. Each shows the full variable dictionary plus a sortable statistics table with mini distributions and data coverage.

Expand all datasets expand to search (Ctrl/⌘+F) or print across all datasets

Known limitations & caveats

• Net effect, not tax-only. The ATT is the gap between actual Kansas and a synthetic blend; over 2012–2016 Kansas also faced a severe drought and aerospace-sector shocks that SCM cannot strip out separately.
• Annual revenue series. rev_state_total, rev_local_total, revenuepop, and the per-capita revstatecapita/revlocalcapita are recorded only once a year and are NA in the other three quarters (2,250–2,850 missing of 5,250).
• Single treated unit, modest donor pool. One Kansas and 49 donors limit statistical power; the four inference methods disagree at the margin (jackknife+ excludes zero; conformal p = 0.066; permutation p = 0.10).
• SCM identifying assumption. Results rest on a convex (or lightly extrapolated) blend of donors standing in for Kansas's untreated path, and on no donor state being affected by Kansas's policy.