Carbon Taxes and CO2 Emissions: Sweden as a Case Study

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/python_sc_co2tax/data/"
use "${BASE}carbontax_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/python_sc_co2tax/data/"
df = pd.read_stata(BASE + "carbontax_data.dta")

# load every dataset at once
files = ["carbontax_data", "disentangling_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 + "carbontax_data.dta", "carbontax_data.dta")
df, meta = pyreadstat.read_dta("carbontax_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/python_sc_co2tax/data/"
df <- read_dta(paste0(BASE, "carbontax_data.dta"))

Overview & sources

Companion data for a Python tutorial that replicates Andersson (2019, AEJ: Economic Policy) on whether Sweden's 1991 carbon tax cut transport CO₂ emissions — and at what economic cost. The headline dataset is an OECD panel of 15 advanced economies observed over 1960–2005 (46 years; 30 pre-treatment, 16 post-treatment), measuring per-capita CO₂ emissions from transport in metric tons alongside the predictors used to build the counterfactual (GDP per capita, vehicles, gasoline consumption, urbanisation, population density). The post layers a naive before/after comparison, difference-in-differences, and the synthetic-control method (pysyncon) — validated by in-time, in-space, and leave-one-out placebo tests — then OLS/IV demand regressions (pyfixest). Synthetic Sweden, built from six donors (Denmark, Belgium, New Zealand, Greece, United States, Switzerland), implies an average 11.3% annual reduction over 1990–2005 with no measurable growth penalty. A second file carries the three counterfactual emission paths used to disentangle the carbon tax from the bundled VAT.

Data sources

Cite this data

Please cite this dataset as follows.

APA

Mendez, C. (2026). Carbon Taxes and CO2 Emissions: A Synthetic-Control Analysis in Python [Data set]. https://carlos-mendez.org/post/python_sc_co2tax/

Andersson, J. J. (2019). Carbon Taxes and CO2 Emissions: Sweden as a Case Study. American Economic Journal: Economic Policy, 11(4), 1–30. https://doi.org/10.1257/pol.20170144 — 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.

BibTeX

@misc{mendez2026pythonscco2tax,
  author       = {Mendez, Carlos},
  title        = {Carbon Taxes and CO2 Emissions: A Synthetic-Control Analysis in Python},
  year         = {2026},
  howpublished = {\url{https://carlos-mendez.org/post/python_sc_co2tax/}},
  note         = {Data set}
}

@article{andersson2019carbon,
  author  = {Andersson, Julius J.},
  title   = {Carbon Taxes and {CO2} Emissions: Sweden as a Case Study},
  journal = {American Economic Journal: Economic Policy},
  volume  = {11}, number = {4}, pages = {1--30}, year = {2019},
  doi     = {10.1257/pol.20170144}
}
@article{abadie2010synthetic,
  author  = {Abadie, Alberto and Diamond, Alexis and Hainmueller, Jens},
  title   = {Synthetic Control Methods for Comparative Case Studies},
  journal = {Journal of the American Statistical Association},
  volume  = {105}, number = {490}, pages = {493--505}, year = {2010}
}

Variable explorer search & filter all 14 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 outcome throughout is per-capita CO₂ emissions from transport (CO2_transport_capita), in metric tons per person per year.

Synthetic control. Let X₁ be the pre-treatment predictor vector for Sweden and X₀ the same predictors for the donor countries (one column per donor). The donor weights w minimise the weighted distance

• w* = argminₙ (X₁ − X₀w)ᵀ V (X₁ − X₀w) subject to w ₃ ≥ 0 and Σ ₃ w₃ = 1 — a convex combination, no extrapolation.
• The diagonal matrix V (predictor importances) is itself chosen to minimise the pre-treatment mean squared prediction error (MSPE) of the outcome, CO₂.
• Predictors: GDP_per_capita, vehicles_capita, gas_cons_capita, urban_pop (means over 1980–1989), plus three lagged outcome levels (CO₂ in 1970, 1980, 1989).

Treatment gap: year-by-year gapₜ = CO2ₙₜ(Sweden) − CO2ₙₜ(Synthetic Sweden), where the synthetic series is Σ ₃ w₃ · CO2ₙₜ(donor j).

Permutation inference: re-run the SCM treating each unit as treated; rank by the post/pre MSPE ratio MSPEₙₒₛₜ / MSPEₙₕₑ; the p-value is the share of units with a ratio ≥ Sweden's (with 15 units the floor is 1/15 ≈ 0.067).

Disentangling. The three columns in disentangling_data are simulated emission paths under different tax bundles. The vertical gap between two paths attributes the CO₂ reduction to the component switched between them — e.g. (NoCarbonTaxWithVAT − CarbonTaxandVAT) / NoCarbonTaxWithVAT is the carbon-tax-only share.

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

• Single-country case. Sweden is one treated unit; external validity to larger emitters or developing economies is not guaranteed.
• Donor-pool size caps the p-value. With 15 countries the smallest possible permutation p-value is 1/15 ≈ 0.067 — exactly the value the post hits.
• Coverage gaps. GDP_per_capita is missing for 10 of 690 country-year rows (688 present); the synthetic-control predictors are averaged over windows where data exist.
• Simulated counterfactual paths. The three *CarbonTax*/*VAT* columns in disentangling_data are model-simulated emission scenarios (defined only from 1970 onward), not raw observations; they normalise differently from the synthetic-control baseline, so the carbon-tax-only share (≈9.5% here) and Andersson's headline (6.3%) describe the same physical wedge under different denominators.
• Window ends in 2005. The panel predates the electric-vehicle surge and later EU climate policy; re-running with newer data would test persistence.