Bayesian Spatial Synthetic Control: California's Proposition 99

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_sc_bayes_spatial/data/"
use "${BASE}r_sc_bayes_spatial_source_data.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_sc_bayes_spatial/data/"
df = pd.read_stata(BASE + "r_sc_bayes_spatial_source_data.dta")

# load every dataset at once
files = ["r_sc_bayes_spatial_source_data"]
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 + "r_sc_bayes_spatial_source_data.dta", "r_sc_bayes_spatial_source_data.dta")
df, meta = pyreadstat.read_dta("r_sc_bayes_spatial_source_data.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_sc_bayes_spatial/data/"
df <- read_dta(paste0(BASE, "r_sc_bayes_spatial_source_data.dta"))

Overview & sources

Companion data for an R tutorial that replicates the California case study of Sakaguchi & Tagawa (2026) on cigarette consumption and Proposition 99. The single file is the balanced state-year panel bundled in the scspill replication package — the same real tobacco panel introduced by Abadie, Diamond & Hainmueller (2010): per-capita cigarette sales (cigsale) and real retail price (retprice) for 39 US states over 1970–2000. California is the one treated unit (Prop 99 switches on in 1988); the other 38 states form the donor pool. The post fits three nested synthetic-control estimators on this panel — classical SCM (tidysynth), a Bayesian horseshoe-prior SCM, and a Bayesian spatial SCM with a spatial autoregressive (SAR) layer — and reads the ATT, the donor weights, and the per-state spillovers off each.

Data sources

Cite this data

Please cite this dataset as follows.

APA

Mendez, C. (2026). Bayesian Spatial Synthetic Control: California's Proposition 99 in R [Data set]. https://carlos-mendez.org/post/r_sc_bayes_spatial/

Sakaguchi, S., & Tagawa, H. (2026). Identification and Bayesian Inference for Synthetic Control Methods with Spillover Effects. The Econometrics Journal. https://doi.org/10.1093/ectj/utag006
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. https://doi.org/10.1198/jasa.2009.ap08746

BibTeX

@misc{mendez2026rscbayesspatial,
  author       = {Mendez, Carlos},
  title        = {Bayesian Spatial Synthetic Control: California's Proposition 99 in R},
  year         = {2026},
  howpublished = {\url{https://carlos-mendez.org/post/r_sc_bayes_spatial/}},
  note         = {Data set}
}

@article{sakaguchi2026spillover,
  author  = {Sakaguchi, Shosei and Tagawa, Hisahiro},
  title   = {Identification and {Bayesian} Inference for Synthetic Control Methods with Spillover Effects},
  journal = {The Econometrics Journal},
  year    = {2026},
  doi     = {10.1093/ectj/utag006}
}
@article{abadie2010synthetic,
  author  = {Abadie, Alberto and Diamond, Alexis and Hainmueller, Jens},
  title   = {Synthetic Control Methods for Comparative Case Studies: Estimating the Effect of California's Tobacco Control Program},
  journal = {Journal of the American Statistical Association},
  volume  = {105}, number = {490}, pages = {493--505}, year = {2010},
  doi     = {10.1198/jasa.2009.ap08746}
}

Variable explorer search & filter all 6 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 data are an observed state-year panel; the post derives no new columns — the three estimators below all operate on the same cigsale outcome. California is the treated unit; the 38 other states are donors.

• ATT: ATT = mean_t( Y_obs(CA, t) − Y_synth(CA, t) ) averaged over the post-treatment years 1988–2000, where the synthetic California is a weighted blend of donor states.
• Classical SCM (Stage 1): choose donor weights on the simplex (α_j ≥ 0, Σ α_j = 1) that minimize pre-treatment fit error, via tidysynth.
• Bayesian horseshoe SCM (Stage 2): α_j ~ N(0, τ² λ_j²) with τ, λ_j ~ HalfCauchy(0, 1) — heavy-tailed shrinkage on the same weights, relaxing the simplex.
• Bayesian spatial SCM (Stage 3): Y_c,t = ρ·W·Y_c,t + X_c,t·β + Y_c^lag·α + ε_t — adds a SAR layer with contiguity matrix W and spatial autocorrelation ρ, relaxing SUTVA, so per-donor spillovers become estimable.

The treatment dummy is constructed as 1 if state == "California" & year ≥ 1988 else 0 (the package's Prop 99 convention). Everything else in the file is observed data read directly from the replication package.

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

• Narrow predictor set. The shipped panel carries only cigsale and retprice — not the log income, youth-share, or beer-sales predictors Abadie (2010) used. This is the main reason the classical ATT (≈ −18) is smaller in magnitude than Abadie's published ≈ −27.
• Spatial weights live elsewhere. The SAR layer needs California's contiguity vector w and the 38×38 donor contiguity matrix W, which ship inside the scspill package, not in this CSV. The data dictionary documents only the outcome panel.
• Tutorial-scale MCMC. The post runs 5,000 iterations (vs the paper's 100,000), so the SAR ρ effective sample size is only ~3 and the Stage 3 credible interval is illustrative, not paper-grade. This is an analysis caveat, not a data caveat — the panel itself is exact.
• Treatment timing. Prop 99 was approved in November 1988 and took effect January 1989, but the package convention codes 1988 as the first post-treatment year; the treatment dummy follows that convention.